Executive function and the dysexecutive syndrome Masud Husain

Dept Experimental Psychology & Nuffield Dept Clinical Neurosciences, University of Oxford

1 What are ? They’re functions that are thought to be deployed when control needs to be exerted

§ Typically described as ‘supervisory’ or ‘controlling’ § Deployed when a situation is novel or difficult § When you need to pay attention because there isn’t an automatic / habitual response to the problem or the automatic response would be inappropriate Example: When the phone rings in someone else’s house, you don’t pick it up, even though the automatic tendency in your own home is to do so § When several cognitive processes need to be co-ordinated § Or when you need to shift from one type of process to another

2 Orchestration of behaviour They’re functions that are thought to be deployed when control needs to be exerted

§ Initiate § Maintain / Sustain / Invigorate / Energise § Stop ongoing action § Inhibit inappropriate behaviour or prepotent response § Switch to a different behavioural set / set shifting / mental flexibility § : manipulation of items in short term memory § Monitor consequences of behaviour / error monitoring § Planning and prioritization § Multi-tasking § Social / emotional engagement

3 When executive function breaks down There might be profound consequences

Executive function Associated Clinical presentation

Task initiation Reduced self-generated behaviours Akinetic mutism, abulia, apathy Procrastination Reduced fluency Maintain / sustain actions Poor ability to stay on task or sustain attention Distractible

Response inhibition Difficulty inhibiting behaviours Disinhibited Acting ’without thinking’ Cognitive flexibility Rigid thinking / stuck on thoughts, concepts or tasks Inflexible / perseverative Shifting behavioural set Working memory Difficulty holding information ”on-line” required to perform Difficulty following instructions tasks or solving problems Self monitoring Lack of awareness when errors or inappropriate responses Unaware are made Planning and prioritization Poor planning / organizational skills Difficulty making decisions Inefficient use of time Multi-tasking Difficulty co-ordinating activities simultaneously or Difficulty solving tasks that sequentially require multi-tasking Social / Emotional Poor social skills, emotional lability, frustration with self / Socially and/or emotionally 4 engagement / Theory of Mind others inappropriate Executive function tests A wide range of tests have been developed to measure executive functions

Executive function Cognitive measure Example

Task initiation Verbal fluency Words beginning with F; designs Non-verbal fluency joining four dots on a grid in a minute Maintain / sustain actions Sustained attention Continuous performance test

Response inhibition Inhibition of pre-potent response Go / No Go; Stroop; Hayling tests

Cognitive flexibility Set shifting Wisconsin card sort test (WCST) Shifting behavioural set Switch cost Trail making B Working memory Verbal working memory forwards and backwards Digit span Visuospatial working memory forwards and backwards Corsi blocks Self monitoring Error detection and correction Perseveration on WCST

Planning Planning and problem solving Tower of London / Tower of Hanoi tests Multi-tasking and Optimal allocation of time Multiple errands; six elements prioritization tests Social / Emotional Ability to infer the thoughts of others Tests of theory of mind 5 engagement / Theory of Mind What does an executive function test? Does it test one process or are there potentially many?

Wisconsin Card Sort Test Patients have to generate new options for sorting the cards (by different possible rules)

But they also have to monitor performance and inhibit themselves from perseverating on a rule they have already used to sort the cards

6 What does an executive function test? Does it test one process or are there potentially many?

Tower of London Test Patients have to generate solutions to problems of different levels of complexity (optimal solutions of 2,4 or 5 moves)

But they also have to monitor their intended plans as well as their execution of plans

If a plan does not seem to provide a correct solution, they have to shift to a different plan

7 Frontal damage and its consequences The modern Phineas Gage

1937: University student, aged 21, during Spanish Civil War, suffered an injury while attempting to escape down a drainpipe which gave way.

Subsequently unable to work on simple manual jobs without supervision.

Required help with everyday tasks. Difficulties managing money.

Apathetic and had problems with initiating, continuing and finishing tasks.

Restless and impatient. “Always cheerful”. No antisocial behaviour reported. 8 Mataro et al (2001) Arch Neurol Frontal damage and its consequences The modern Phineas Gage

9 Mataro et al (2001) Arch Neurol Frontal damage and its consequences The modern Phineas Gage

10 Mataro et al (2001) Arch Neurol Frontal damage and its consequences The modern Phineas Gage

11 Mataro et al (2001) Arch Neurol Failure on an executive function test Doesn’t mean the patient has a primary executive function deficit

Wisconsin Card Sort Test Nor is it the case that patients who have difficulties that seem to suggest a dysexecutive syndrome in real life always show deficits on executive function tests

12 Frontal damage and its consequences Eslinger & Damasio’s case EVR

§ High-achieving young man – chief accountant by age 29 § Aged 35: Personality changes. Large orbitofrontal meningioma resected § Behavioural change: invested all his savings in a risky venture, became bankrupt § Drifted through different jobs including warehouse labourer, but fired from each § Employers complained about poor time keeping and disorganization § Wife left with children and divorced him after 17 yrs marriage. He moved in with his parents § A month after divorce he remarried aganst advice of relatives and divorced 2 yrs later § Needed 2hrs to get ready for work. Some days consumed by shaving and washing his hair § If he was planning to have dinner out, planning took hours deciding seating plan, menu, etc

13 Eslinger & Damasio (1985) Neurology Frontal damage and its consequences EVR passed standard executive function tests that were available at the time, yet clearly was not normal Key finding: Dissociation between intact cognitive abilities using standard tests (e.g., Wisconsin card sort) and poor use of cognition in real world environments (e.g. planning a dinner out).

Figure 2. Sagittal MRI of EVR obtained with inversion recovery technique. The incidence 14 A and level of the cuts are shown in the lower right Figure 1. CT images and templates of patient EVR. Each cut was 8 mm. There was a low-density area in both corner. InEslinger the right & Damasio (1985) Neurology frontal lobes, which merged with the image of the ventricular system, corresponding to the surgical resection. The hemisphere, the large area resection involved (I) all of the orbital cortex on the right and part of the orbital cortex on the left; this lesion of low signal involved the included both frontal poles and corresponded on the right to Brodmnnnk cytoarchitectonic fields 11, 12, and 25; on orbital and prefrontal the left, field 25 is intact and fie& 11 and 12 were only partially damaged; white mutter subjacent to these orbital cortices as well as underlying white matter. I It also extended into premotor areas. In the left flow) and through the orbital . Slices 2, 3, epilepsy (table 3). Premorbid intelligence and person- hemisphere, the area of and 4 showed a large region of low flow in both frontal ality were normal in all cases. All patients were studied low signal was more lobes, especially so in the right hemisphere, correspond- in our unit with comparable behavioral and anatomic restricted, sparing some ing to the ablation. The left dorsolateral frontal lobe, methodologies. orbital cortex but involving the basal forebrain region, the and underlying white matter. thalamus, the rolandic cortices, and all temporal, pari- Discussion. The findings from EVR are of special etal, and occipital cortices had levels of cerebral blood importance for at least three reasons. First, they pro- flow comparable with normal controls. vide insight into cognitive mechanisms of sociopathic Controls. All controls had nonpsychiatric conditions behavior. Second, they add to our understanding of caused by stroke or herpes simplex encephalitis, and no orbital and lower mesial frontal lobe function.1736 Third, NEUROLOGY 35 December 1985

1734 NEUROLOGY 35 December 1985 FRONTAL LOBE DAMAGE 733

TABLE 4. PERFORMANCE ON THE SIX ELEMENT TASK Max. time on any sublask Case No. of subtasks tackled (min, s) 1 (I) 2 7,30 1 (II) 4 6,19 2(1) 5 10,11 2 (II) 3(+2 incorrect) 6,22 3 (I) 3 7,18 3 (II) 2 9,30 Controls 5.7 (±0.5) 5,35 (±0,53)

Results. Two measures of task performance are shown in Table 4. Two different attempts were made by each patient, at least 3 wks apart. On the first attempt Cases 1 and 3 tackled less subtasks than any control and all 3 spent more time on their subtask they attempted for longest (see Table 4). Qualitatively the behaviour of the patients was also atypical. On his first attempt A.P. made notes for over 4 min to help with the dictation task, but in fact never dictated at all. He only attempted 2 subtasks, the second of these occupying the last 7.5 min. On his second attempt he tackled only 4 of the subtasks claiming that he did not dictate his journey back as he had not decided where he would go when he would leave the hospital! Case 2 succeeded in tackling 5 subtasks but spent much longer than any normal subject— over 10 min in 3 separate periods—on 1 of the subtasks (maths I) without ever trying its complementary task (maths II). On his second attempt he behaved bizarrely, changing task 63 times (mean task changes of controls 5.7, SD 2.19) and dictating the two sets of picture names—spending only 14 s and 20 s on them—instead of writing them down; no normal subject carried out a subtask incorrectly. Also on 6 occasions he carried out a subtask immediately after the complementary one, so breaking one of the rules. (One control also made this error. She said she had mistaken the rule as not doing a part I of a pair of subtasks after another part I; this she correctly applied.) Case 3, on her first attempt, only tackled 3 subtasks. On the second she wrote her journey instead of dictating it, attempted only 2 other subtasks and spent the last 9 min on just 1 of them. During this time she looked at the stopwatch on 7 occasions but did not switch tasks. On the 2 timing tasks Case 1 performed well on both (1.5 min, stopped 1 min 17 s; 2.5 min, stopped 2 min 28 s). Case 2, however, failed the first, stopping at 1 min 38 s although being fully aware he had to stop at 1 min 15 s. On the second task his performance was very poor on both testing occasions. On the first attempt he stopped after 1 min 37 s, saying that he thought 2 min 45 s had elapsed and on his second he failed to stop until 3 min 26 s had elapsed. Case 3 performed well on the 1.25 min task (stopping at 1 min 15 s). However, on the more difficult 2.5 min task she did not stop until 3 min 20 s had passed, despite having looked at the watch at 2 min 38 s. No control was more than 1 s out on the first task or more than 5 s out on the second. (Mean time out for task 1, 0.1 s, SD 0.3; task 2, mean 2.9 s, SD 2.0).

Comment Ecologically valid executive function testsAll 3 patients performed at below the normal range on quantitative measures of performance on the Six Element task. In addition their performance was also qualitatively atypical. Multiple Errands Test (ME). The purpose of this test, which is undertaken in a pedestrian precinct near Developed in an attempt to capture the difficulties patients seem tothe hospitaencounterl previously unknow inn to threale patients life, is for them to carry out a number of tasks in situations Ecologically valid executive function tests 203 where minor unforeseen events can occur. The subtasks the patients had to carry out are basically very simple except for one, which has subcomponents designed to be reasonably demanding for someone of the IQ level and cultural background of the patients. While still inside the hospital the patient is given a card with 8 tasks written on it, 6 of which are simple (e.g., buy a brown loaf, buy a packet of throat pastilles). A seventh requires the subject to be at a certain place 15 min after starting. An eighth is more demanding, 4 sets of information have to be obtained and written on a postcard, namely: (1) the name of the shop in the street likely to have the most expensive item; (2) the price of a pound of tomatoes; (3) the name of the coldest place in Britain yesterday; and (4) the rate of the exchange of the French franc yesterday. 734 T. SHALLICE AND P. W. BURGESS

The card also contains the following instructions: 'You are to spend as little money as possible (within reason) and take as little time as possible (without rushing excessively). No shop should be entered other than to buy something. Please tell one or other of us when you leave a shop what you have bought. You Downloaded from arhttps://academic.oup.com/brain/article-abstract/114/2/727/263651e not to use anything not bought on the street (other than a watch) to assist you. You may do the tasks by The Librarian. user on 05 November 2017in any order.' This second part of the instructions—the 'rules'—is then read aloud to the patient who is asked to repeat them. If the rules cannot all be repeated satisfactorily the procedure is repeated until they can be. The subject is then taken the short distance to a shopping precinct where he or she is asked to repeat the rules again. If there is any failure to reproduce the gist of any instruction the subject is reminded of it. It is then indicated to the patient where the limits of the test area are; they are very clearly demarcated by a large street ending in the pedestrian precinct at one end and a set of traffic lights at the other. The behaviour of the subjects while carrying out the activity is monitored by 2 observers and they are also debriefed after the task had been completed. Controls. Nine control subjects were tested. They were matched with the patients for age (mean 40, range 24-63 yrs) and NART IQ (mean 122, range 113-127). Results. Table 5 gives the number of errors produced by each of the 3 patients. Each produced at least 2 SDs more than the controls. Table 5 also shows the errors broken down into subcategories: (1) inefficiencies—where a more effective strategy could have been applied, e.g., entering the same shop more than once; (2) rule breaks—where a specific rule (either social or explicitly mentioned in the task) is broken, e.g., going outside the boundaries or leaving a shop with a newspaper without paying; (3) interpretation failure—where the requirements of a particular task are misunderstood, e.g., assuming that the information must be written on the birthday card rather than the postcard; (4) task failure—a task either not carried out or not completed satisfactorily.

TABLE 5. ERRORS ON MULTIPLE ERRANDS Cases / 2 3 Control Inefficiencies 6* 9* 5* 1.4(1.1) Rule breaks 5* 8* 8* 1.6(1.3) Interpretation failures 1 1 1 0.4 (0.7) 15 (a) (b) Task failures 0 5* 4* 1.1 (1.4) Total errors 12* 23* 17* 4.6(2.1) Fig. 2. Panel A shows a schematic representation of a typical control performance on Shallice and Burgess’sDepiction (1991a) of Multiple Errands Test performance described in Shallice & Burgess (1991) Brain Multiple Errands Test. Panel B shows the impaired performance of a patient (DN) with frontal lobe damage who was * More than 2 SD worse than the control subjects. matched to the control for intellectual functioning, and who also performed well on ten traditional clinical tests of All 3 patients performed at the 5% level or worse, as estimated from the control data, on both the number executive function (see Shallice & Burgess 1991a for further details). Hatched shading indicates shops that need to be of inefficiencies and on the number of rule breaks. There was no difference overall in the rate of errors entered in order to complete the test. Solid block shading indicates shops that do not need to be entered, or are of the other two types, although Case 3 produced more task failures. Thus 2 normal controls as well as forbidden by the task rules. (Adapted from Burgess & Alderman, 2004.) the patients failed to provide the required information on the postcard, probably because the information required was placed on the back of the instruction sheet. Qualitatively the performance of the patients was even worse. They made types of error which were tive function” clinical tests. Certainly the one above yielded Moreover, the study of differences in patterns of failure not produced by any control. Case 3 used a clearly irrelevant criterion inappropriately when she broke a rule (entering a shop without buying anything) because when in the shop (a chemist) she found it did aclinicaltest(theSixElementTest)whichiswidelyused, on the two tests are revealing patterns of impairment that not have a soap she especially liked; other cheap soap—which would have been at least as adequate for and also a useful task with high “representativeness” (the are both theoretically interesting and clinically relevant (see the task— was available. She also failed to note the time when starting. Cases 1 and 2 both became involved in social complications. Case 1 made the interpretative error of deciding he needed yesterday's newspaper MET). There are now various versions of both the Multiple Burgess et al., 2005b for further details). For instance, Alder- (to find the coldest place in Britain on the previous day). He entered a newsagents and asked if they had Errands (e.g., Knight et al., 2002) and the Six Element tests man et al. (2003) found a double-dissociation between rule- the previous day's paper. They had, so he walked out with it, incidentally breaking the buying rule. He was pursued into the street by the shopkeeper who wanted to be paid for it. A.P. had just assumed that (e.g., Burgess et al., 1996b; Emslie et al., 2003; Kliegel breaking behaviour and failure to initiate tasks, and these because a previous day's newspapers are generally worthless he could have one without paying. Case 2 et al., 2000; Levine et al., 2000; Manly et al., 2002), and two patterns were associated with different dysexecutive produced a complex set of errors. Thus one series of actions began with his being the only subject to enter a shop simply to ask about the location of another type of shop (breaking buying rule). He was referred they have been used in the investigation of a variety of symptoms in everyday life. “Rule-breakers” tend to show to a shop outside the allowed area, went to that shop (breaking the limit rule) and became involved in conditions (e.g., traumatic brain injury, Hoclet et al., 2003; problems with control aspects of memory (e.g., show con- Levine et al., 2000; , Evans et al., 1997; depres- fabulation, temporal sequencing problems, and repetitive sion, Channon & Green, 1999; ADHD and Oppositional behaviour). By contrast, people who fail to complete the set Defiant0Conduct Disorder, Clark et al., 2000; drug abuse, tasks tend to show negative symptoms, such as apathy and Zakzanis & Young, 2001; sleep deprivation, Nilsson et al., shallow affect. From a theoretical perspective, BurgessDownloaded et al. from https://academic.oup.com/brain/article-abstract/114/2/727/263651 by The Librarian. user 2005; normal ageing, Garden et al., 2001; “intensive care (2005b) argue that these results may reflect theon 05 “real- November 2017 syndrome,” Sukantarat et al., in press), and there is also world” consequences of a time0event dissociation in pro- information about the psychometric properties of the tests, spective memory. From a rehabilitation perspective the (e.g., Alderman et al., 2003, Jelicic et al., 2001; Kafer & patterns suggest different treatment applications (Burgess Hunter, 1997; Knight et al., 2002), their “generalisability” &Robertson,2002)inamuchmoredirectwaythanwould, (Burgess et al., 1998), and their correlations with other exec- say, a dissociation between perseverative behaviour and prob- utive function tests (Duncan et al., 1997). lems with concept shifting in performance on the WCST. Consequences of frontal damage Our case with bilateral orbitofrontal / ventromedial prefrontal contusions

§ Partner describes inappropriate behaviour, bouts of anger, lack of empathy and disinhibition

§ He may hug men and women he meets for the first time, be over-familiar with them or hog the conversation, not waiting his turn to speak

§ He says: "I do not get it. I just do not twig that what I am saying is offensive!”

§ He shows difficulty in making choices, e.g. when picking from a restaurant menu

16 Consequences of frontal damage | Patient video

17 620 | Brain 2009: 132; 617–627 S. G. Shamay-Tsoory et al.

Table 1 Patients demographic details and performance on measurements of executive functions, RAVEN and BDI

VM ( N = 11) IFG ( N = 8) PC ( N = 11) Male = 9 Male = 8 Male = 7 Female = 2 Female = 0 Female = 4 Age, mean (SD) 36.45 (16.20) 32.75 (15.06) 38.00 (14.89) Years of education, mean (SD) 11.70 (1.41) 14.12 (2.58) 13.36 (1.74) Time since injury in years (SD) 9.36 (11.85) 7.25 (6.94) 7.27 (5.38) Laterality Right = 3 Right = 5 Right = 3 Left = 8 Left = 3 Left = 8 Bilateral = 3 Etiology Head injury = 8 Head injury = 6 Head injury = 6 Tumor = 2 Tumor = 2 Tumor = 3 Stroke = 1 Stroke = 2 WCST set loss, mean (SD) 0.54 (1.21) 1.0 (1.26) 0.90 (1.19) WCST perseverative errors, mean (SD) 11.27 (7.146) 15.67 (4.62) 13.10 (7.40) WCST total errors, mean (SD) 22.90 (12.38) 31.33 (3.93) 23.80 (13.18) Digit span, mean (SD) 8.22 (1.64) 8.00 (1.58) 7.88 (1.86) Fluency phonemic, mean (SD) 9.18 (2.85) 12.25 (7.19) 12.00 (4.96) Fluency semantic, mean (SD) 17.86 (5.44) 22.25 (7.19) 19.54 (5.56) Raven, mean (SD) 39.40 (28.32) 40.00 (22.32) 60.25 (26.55) Similarities, mean (SD) 9.50 (0.79) 11.40 (2.07) 11.00 (2.13) BDI, mean (SD) 18.25 (8.68) 13.50 (11.36) 12.36 (7.21)

in imaging. Correlation of lesion volumes between experimenters perspective-taking scale (PT) which measures the reported tendency was highly significant (r = 0.749, P = 0.005), indicating the high relia- to adopt spontaneously the psychological point of view of others (‘I bility620 of| thisBrain method 2009: across 132; experimenters. 617–627 sometimes try to understand my friendsS. better G. Shamay-Tsoory by imagininget how al. Subjects were then divided into three groups according to Damasio things looks from their perspective’); (ii) the fantasy scale (FS), mea- and Damasio (1989): the VM group, if damage involved mostly the suring the tendency to imaginatively transpose oneself into fictional Table 1 Patients demographic details and performance on measurements of executive functions, RAVEN and BDI orbitofrontal and/or the ventral portion of the medial wall of the situations (‘When I am reading an interesting story or novel I imagine frontal lobe (BAs: mesial 8 and 9, 10, 24 and 32, 10, 11 and 47); VM ( N = 11)how I would feel if IFG the ( eventsN = 8) in the story were happening PC ( N = to 11) me’). the IFG group, if damage involved mostly the Pars opercularis and the The PT was found to be consistently related to measures of interper- Social / emotional engagementPars triangularis (BAs 44, 45), and a group of patients withMale posterior = 9sonal functioning, social Male =competence 8 and high self esteem Male= but 7 not to 624 | Brain 2009: 132; 617–627 lesion involving damage outside the frontal lobes comprisingS.Female one G. =of 2 Shamay-Tsooryaffective empathy Female(Davis,et = 1983). 0 al. The FS is highly related Female to = 4 verbal theAge, two mean control (SD) groups. Figure 1 presents lesion superimposition 36.45 (16.20)for measures and intellectual 32.75 (15.06) abilities, especially verbal intelligence. 38.00 (14.89) Both Emotional versus cognitive empathy theYears VM, of IFG education, and PC groups mean(SD) (lesions are flipped to one hemisphere 11.70 (1.41) to scales were found to 14.12 be positively (2.58) correlated with other 13.36 validated (1.74) mea- enhanceTime since anatomical injury in overlap). years (SD) In two cases, patients assigned 9.36 to (11.85) the sures of cognitive empathy 7.25 (6.94) such as the Hogan empathy 7.27 scale (5.38) (Hogan, IFGLaterality group had damage that extended to include portions of Rightarea 6; = 3 in 1969), suggesting that Right these = 5 scales indeed measure cognitive Right = 3empathy four cases lesions involved also area 48 and in three patientsLeft = 8 the (Davis, 1983). Left = 3 Left = 8 damage also included the temporal pole (BA 38) extendingBilateral to the = 3 The two affective scales are the EC and the personal distress scale beginningEtiology of area 47. Among patients assigned to the VM Head group, injury = 8(PD). The EC scale tapsHead the injury respondents’ = 6 feelings of warmth, Head injury compas- = 6 four had a lesion extending to BA 46, and in one patient theTumor lesion = 2sion and concern forTumor others = 2 (e.g. ‘I often have tender, Tumor concerned = 3 reached BA 25. While most of the patients had unilateral lesions,Stroke three = 1feelings for people less fortunate than me’). The PD Stroke scale = assesses2 patientsWCST from set loss, the mean posterior (SD) lesion control group had bilateral 0.54 lesions. (1.21)self-oriented feelings 1.0 of (1.26) anxiety and discomfort resulting 0.90 from(1.19) tense 3 3 TheWCST volume perseverative of lesions errors, ranged mean from (SD) 0.90 cm to 145.317 11.27 cm (7.146)interpersonal settings 15.67 (e.g. (4.62) ‘being in a tense emotional 13.10 (7.40) situation (mean = 33.71 cm3, SD = 33.03 cm3) and there were no significant dif- WCST total errors, mean (SD) 22.90 (12.38)scares me’). 31.33 (3.93) 23.80 (13.18) ferences in the size of lesions among the three lesion groups Digit span, mean (SD) 8.22 (1.64)Individual scores 8.00for each (1.58) item (on a scale from 1 to 7.88 5) were (1.86) trans- [F(2,29) = 1.549, NS], indicating that lesion sizes were not different Fluency phonemic, mean (SD) 9.18 (2.85)formed to a –2 to 2-point 12.25 (7.19) scale, therefore the scores of 12.00 each (4.96) sub-scale between groups. Fluency semantic, mean (SD) 17.86 (5.44)ranged between –14 22.25 and (7.19) +14 points. Full-scale scores 19.54 are not (5.56) calcu- Raven, mean (SD) 39.40 (28.32)lated as each scale40.00 has been (22.32) shown to measure a discrete 60.25 component (26.55) ExperimentalSimilarities, mean (SD) measures 9.50 (0.79)of empathy (Davis, 11.40 1983). (2.07) To assess cognitive empathy 11.00 we (2.13) used the BDI, mean (SD) 18.25 (8.68)mean score of the PT 13.50 and (11.36) the FS sub-scales whereas emotional 12.36 (7.21) empa- Emotional and cognitive empathy thy was assessed using the mean score of the EC and the PD sub- To assess empathy multi-dimensionally, we administered the IRI. The scales. IRI (Davis, 1983), is a 28-item self-report questionnaire that measures The factor structure of the IRI was confirmed in a study of bothin imaging. components Correlation of empathy. of lesion To volumes date, it between is the only experimenters published femaleperspective-taking dieticians (n scale= 217) (PT) and which dietetic measures interns the (n reported= 168) (Spraggins tendency measurewas highly that significant allows a(r = multi-dimensional 0.749, P = 0.005), assessment indicating the of high empathy. relia- etto al adopt., 1990). spontaneously The IRI has the good psychological internal consistency, point of with view alpha of others coeffi- (‘I Thebility questionnaire of this method contains across experimenters.four 7-item scales (two cognitive scales cientssometimes ranging try from to understand 0.68 to 0.79 my(Davis, friends 1983;better Christopher by imagininget how al., andSubjects two affective were then scales). divided The into two three cognitive groups according scales are: to Damasio (i) the 1993).things looks from their perspective’); (ii) the fantasy scale (FS), mea- and Damasio (1989): the VM group, if damage involved mostly the suring the tendency to imaginatively transpose oneself into fictional orbitofrontal and/or the ventral portion of the medial wall of the situations (‘When I am reading an interesting story or novel I imagine frontal lobe (BAs: mesial 8 and 9, 10, 24 and 32, 10, 11 and 47); how I would feel if the events in the story were happening to me’). Downloaded from https://academic.oup.com/brain/article-abstract/132/3/617/336907 by Universitythe of IFGOxford group, - Bodleian if damage Library involved user mostly the Pars opercularis and the The PT was found to be consistently related to measures of interper- on 05 November 2017 Pars triangularis (BAs 44, 45), and a group of patients with posterior sonal functioning, social competence and high self esteem but not to lesion involving damage outside the frontal lobes comprising one of affective empathy (Davis, 1983). The FS is highly related to verbal the two control groups. Figure 1 presents lesion superimposition for measures and intellectual abilities, especially verbal intelligence. Both the VM, IFG and PC groups (lesions are flipped to one hemisphere to scales were found to be positively correlated with other validated mea- Figure 5 Location and overlap of brain lesions according to emotional versusenhance cognitive anatomical empathy overlap). In two impairment-groups. cases, patients assigned to the (suresA) of Lesions cognitive empathy of the such as the Hogan empathy scale (Hogan, emotional-empathy-impaired group (n = 6). Four patients had an IFG damageIFG involving group had damage area that 44,extended one to include had portions a VM of area damage 6; in 1969), and suggesting one that had these scales a indeed measure cognitive empathy four cases lesions involved also area 48 and in three patients the (Davis, 1983). damage also included the temporal pole (BA 38) extending to the The two affective scales are the EC and the personal distress scale PC damage. Chi-square analysis revealed that lesions involving area 44 werebeginning significantly of area 47. Among more patients frequent assigned to in the this VM group, group as compared to 18 2 (PD). The EC scale taps the respondents’ feelings of warmth, compas- the non-impaired group [ (1) = 7.071, P = 0.008]. (B) Lesions of the cognitive-empathy-impairedfour had a lesion extending to BA 46, andgroup in one patient (n = the 7): lesion fivesion had and VM concern damage for others (e.g. ‘I often have tender, concerned reached BA 25. While most of the patients had unilateral lesions, three feelings for people lessShamay fortunate-Tsoory than me’).et al The (2009) PD scale Brain assesses involving area 10 and 11, one had an IFG damage and one had a PC damage.patients Chi-square from the posterior analysis lesion control revealed group had bilateral that lesions. lesionsself-oriented involving feelings of area anxiety and discomfort resulting from tense 3 3 2 2 The volume of lesions ranged from 0.90 cm to 145.317 cm interpersonal settings (e.g. ‘being in a tense emotional situation 10 [ (1) = 6.04, P = 0.010] and area 11 [ (1) = 5.185, P = 0.023] were significantly(mean = 33.71 cm more3, SD = 33.03 frequent cm3) and there in were this no significant group dif- as compared to the scares me’). ferences in the size of lesions among the three lesion groups Individual scores for each item (on a scale from 1 to 5) were trans- non-impaired group. [F(2,29) = 1.549, NS], indicating that lesion sizes were not different formed to a –2 to 2-point scale, therefore the scores of each sub-scale between groups. ranged between –14 and +14 points. Full-scale scores are not calcu- lated as each scale has been shown to measure a discrete component Experimental measures of empathy (Davis, 1983). To assess cognitive empathy we used the and emotion recognition. Specifically, BA 44 was found most cri- compatible with a recent report by Nummenmaamean scoreet of the al PT. and (2008) the FS sub-scales whereas emotional empa- Emotional and cognitive empathy thy was assessed using the mean score of the EC and the PD sub- tical for emotional empathy while areas 11 and 10 were found who proposeTo assess empathy that multi-dimensionally, emotional weempathy administered the facilitates IRI. The scales. motor representa- critical for cognitive empathy. Interestingly, area 44, which has tion ofIRI other (Davis, 1983), peoples’ is a 28-item self-report emotions, questionnaire and that measures results inThe more factor structure vigorous of the IRI was confirmed in a study of both components of empathy. To date, it is the only published female dieticians (n = 217) and dietetic interns (n = 168) (Spraggins been reported as cytoarchitectonically homologous to F5 mirroringmeasure of that the allows observed a multi-dimensional bodily assessment and of emotional empathy. et al states., 1990). The than IRI has good cog- internal consistency, with alpha coeffi- The questionnaire contains four 7-item scales (two cognitive scales cients ranging from 0.68 to 0.79 (Davis, 1983; Christopher et al., (Petrides et al., 2005), was identified as a central part of the nitive empathy.and two affective scales). The two cognitive scales are: (i) the 1993). MNS (Rizzolatti, 2005). Thus, although some doubt has been These results provide strong support for the existence of two Downloaded from https://academic.oup.com/brain/article-abstract/132/3/617/336907 by University of Oxford - Bodleian Library user voiced with respect to the role played by MNS in social cognition behaviouralon 05 November 2017 systems for understanding others: an early emotional (Jacob and Jeannerod, 2005), these results present an empirical matching/mirroring system involving the MNS and a more evidence that the MNS is essential for emotional empathy. advanced system for cognitive understanding of mental states, Nonetheless, although area 44 has been clearly related to social involving the VM cortices. Furthermore, the fact that area 44 cognition, it is not a region which is typically associated with the was found most critical for emotional empathy while BA 11 and emotional aspect of social cognition. How can one explain the 10 are necessary for cognitive empathy is consistent with these essential role of BA 44 in emotional empathy observed in the cortices being, in terms of synaptic hierarchy, unimodal and het- present study? First, beyond language related functions, neuroi- eromodal, respectively (Mesulam, 2000). Additionally, areas 10 maging studies have suggested a central role of BA 44, as a neural and 11 differ from area 44 cytoarchitectonically according to the substrate for imitation (Iacoboni et al., 1999, 2001; Nishitani and layering of the cortex (Brodmann, 1909): while among the six Hari, 2000, 2002; Tanaka et al., 2001; Koski et al., 2002; Tanaka layers of the isocortex layer IV is not fully developed in BA 44, and Inui, 2002; Carr et al., 2003; Grezes et al., 2003; Rizzolatti it is fully developed in areas 11 and 10. Therefore, whereas BA 44 and Craighero, 2004). Second, BA 44 have been increasingly is classified as dysgrangular, areas 11 and 10 are considered implicated in emotion recognition tasks such as identification of as granular cortex. This could be taken as indirect evidence for emotional intonation (Wildgruber et al., 2005) and judgement of these parts of area 44 being phylogenetically older than areas 10 facial expressions (Kesler-West et al., 2001). Additionally, as noted and 11. above, activation of BA 44 has been shown recently to occur not It should be noted that most of the patients in this study sus- only with respect to motor actions but to emotion recognition tained a closed head injury. Although cases of apparent diffuse (Carr et al., 2003). Furthermore, Adolphs et al. (2002) have axonal injury were excluded and the proportion of patients with found deficits in emotion recognition from faces and from prosody head injury was matched between the different groups, it is after damage to either the frontal operculum (involving BA 44) or impossible to completely rule out the possibility that in some right somatosensory cortices. Taken together, it appears that of BA patients lesions were more diffuse. 44 is particularly involved in imitation and emotion recognition. It Additionally, it should be acknowledged that attempting to con- may be interesting to speculate whether imitation, the heart of vert stereotaxic coordinates based on MRI/CT scans to BAs based emotional contagion, is even more enhanced when emotional on cytoarchitecture is problematic, and therefore the conclusions social stimuli rather than neutral stimuli are presented. Thus, we regarding cytoarchitectonic differences should be treated with believe that experiencing emotions may encourage and motivate caution, particularly since most of our patients had CT scans imitation and therefore depend on intact MNS. This conclusion is rather than MRI scans.

Downloaded from https://academic.oup.com/brain/article-abstract/132/3/617/336907 by University of Oxford - Bodleian Library user on 05 November 2017 Theory of mind Assessed for example using the Faux pas test

§ Faux pas: Breach of social etiquette or convention § Example: Jill had just moved into a new apartment. Jill went shopping and bought some new curtains for her bedroom. When she had just finished decorating her apartment, her best friend, Lisa, came over. Jill gave her a our of the apartment and asked, “How do you like my bedroom?” “Those curtains are horrible” Lisa said, “I hope you are going to get some new ones!” Q1 Did Lisa know the curtains were new? Q2 Did someone say something they shouldn’t have?

19 For example of findings in frontal lobe patients, see Torralva et al (2009) Brain Magazine R111 Supervisory attentional system model A general model to explain executive functions | Norman and Shallice technological advances over the last forty years, however, have established links between performance of certain paradigms which putatively make demands upon executive processes Supervisory system and the operation of prefrontal cortex, as assessed by human lesion studies, functional neuroimaging, and other methods (such as electrophysiology). There are many such tasks that can be used in work on humans: the Wisconsin Card Sorting Test, the Tower of London test, the Trail-Making Test and the Hayling Sentence Completion Test, to name but a few. What these Sensory information Schemas Response systems tasks generally have in common Current Biology is that they require processing 20 beyond the instantiation of a single set of well- learned or directly cued Figure 1. The ‘Supervisory System’ model of executive functions. associations between stimuli and According to this model, behaviour is controlled by schemas, which may appropriately be trig- responses. This may be, for instance, gered by incoming perceptual information in routine situations; but in non-routine situations, the because they require: overcoming executive functions of the Supervisory System are called upon to modulate their level of activity. the tendency to enact strong The Supervisory System is now thought to be composed of multiple, interacting, sub-systems. stimulus–response associations that This illustration is based particularly on the framework put forward by Norman and Shallice (see Shallice, 1988). However, other frameworks, such as those presented by Duncan (2001) are currently not relevant (‘inhibition’); and Miller and Cohen (2001) have strong similarities. remembering and manipulating information over delay periods, especially in the face of interference of symptoms, which can occur in situations involving well-learned links (‘working memory’); switching combination or singly. For this reason, between particular events in our between two or more alternative many accounts of the organisation environment and particular ways of stimulus– response mappings of the executive system have behaving. However, in other situations, (‘flexibility’); ‘(self-)initiation’, because concentrated on finding principles that such as those involving novelty or there is an absence of external cues might explain, simultaneously, one or where well-learned responses need to to prompt behaviour; development more of these symptoms. For instance, be inhibited, environmental triggering of a novel strategy or approach impulsivity and a failure to pursue is inadequate and a second system (‘strategy application’); or control of goals over long periods of time are is required to modulate the activity novel behavioural sequences over amongst the most common difficulties level of schemas. Norman and Shallice long periods of time (‘multitasking’ or reported after frontal lobe damage. label this the ‘supervisory system’ and ‘prospective memory’, for example). But, in other situations, frontal lobe suggest that it is supported by the Similar progress has been made in damage can cause ‘perseveration’ (an frontal lobes of the brain. work on animals, except that in this inability to switch to a new behaviour In some situations, environmental case work has largely concentrated when the previous one becomes triggers lead to the activation of one on tasks that have the first three inappropriate). How can these schema, but an alternative schema characteristics. It is now clear that the apparently contradictory patterns needs to be selected. In these dynamics of these types of paradigm be reconciled? situations, damage to the supervisory are complex, and that many other Norman and Shallice put forward an system will make it more likely that the brain regions are involved in their influential framework for understanding previously relevant schema, triggered performance. Moreover, the prefrontal executive function that can potentially by environmental events, will continue cortex, especially in humans, probably account for both of these findings to be selected, leading to excessive supports many abilities that have (Figure 1). According to this framework, behavioural rigidity. In other situations, so far received scant attention (for behaviour is governed by sets of where the task in hand is not strongly example, aspects of social behaviour thought or action ‘schemas’. A schema cued by environmental events, a or creativity). But the development is a set of actions or cognitions that reduction in supervisory input may of these procedures has opened up have become very closely associated lead to the triggering of inappropriate the possibility of objective evaluation through practice. These schemas can behaviour by salient objects in the of what once seemed scientifically become activated in two distinct ways. environment, leading to excessive intractable. Consequently, the field is First, they can be triggered by events distractability. Thus, damage to the currently one of the fastest growing in in one’s environment. For example, supervisory system could explain cognitive neuroscience. when driving, the schema to brake excessive rigidity, and also excessive can be triggered by the sight of a red distractibility, both of which have been Theoretical accounts of executive light (if you are an experienced driver). reported to occur following damage to function Environmental triggering of schemas the frontal lobes. Patients with frontal lobe damage can be sufficient to accomplish Various other frameworks for may present with a wide range appropriate behaviour in routine understanding executive function The ‘dysexecutive syndrome’ Some argue that evidence suggests that there is no unitary syndrome but many different syndromes

Phil. Trans. R. Soc. B (2007) 362, 901–915 doi:10.1098/rstb.2007.2096 Published online 3 April 2007

Is there a dysexecutive syndrome? Donald T. Stuss1,2,* and Michael P. Alexander1,3 1Rotman Research Institute, Baycrest, 3560 Bathurst Street, Toronto, Ontario, Canada M6A 2E1 2Departments of Psychology and Medicine (Neurology, Rehabilitation Sciences), University of Toronto, Toronto, Ontario, Canada M5S 3G3 3Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215, USA The role of the frontal lobes has often been described as a ‘paradox’ or a ‘riddle’. Ascribed to this region has been the loftiest of functions (e.g. executive; seat of wisdom); others contested that the frontal lobes played no special role. There has also been controversy about the unity or diversity of functions related to the frontal lobes. Based on the analysis of the effects of lesions of the frontal lobes, we propose that there are discrete categories of functions within the frontal lobes, of which ‘executive’ functioning is one. Within the executive category, the data do not support the concept of an undifferentiated central executive/supervisory system. The results are better explained as impairments in a collection of anatomically and functionally independent but interrelated processes. Evidence for three separate frontal attentional processes is presented. For each process, we present an operational description, the data supporting the distinctiveness of each 21 process and the evidence for impairments of each process after lesions in specific frontal regions. These processes and their coarse frontal localizations are energization—superior medial, task setting— left lateral and monitoring—right lateral. The strength of the findings lies in replication: across different tasks; across different cognitive modalities (e.g. reaction time paradigms, memory); and across different patient groups. This convergence minimizes the possibility that any of the findings are limited to a specific task or to a specific set of patients. Although distinct, these processes are flexibly assembled in response to context, complexity and intention over real time into different networks within the frontal regions and between frontal and posterior regions. Keywords: frontal lobes; dysexecutive syndrome; attention; monitoring; energization; task setting

1. INTRODUCTION functions, and both may be recruited for complex (a) Evolution of the question and initial response tasks—gambling decision, investment planning, etc. Their relatively large size, late evolutionary development Lesions may disrupt either or both, depending on site. It and rich anatomical connectivity all strongly suggest a may be the interaction of emotional status and cognition central role, or roles, for the frontal lobes in human that determines many behaviours, but it is the cognitive cognition and emotion. That there are so many competing aspect of tasks that are defined by executive functions. theories about frontal functions despite the many recent Many prominent theoretical positions emphasized the advances in lesion and imaging research illuminates the dominant role of the frontal lobes in organizing cognition, difficulty in understanding this complex region. thus such terms as supervisory system and central executive. The difficulties in studying the frontal lobes are A controversy within this approach has been related myriad. First, there is no particular predisposition for a to the unity (Duncan & Miller 2002) versus diversity neurological disorder to the frontal lobes. Although (Stuss & Benson 1986; Shallice 2002) of executive cerebrovascular disorders may damage only the frontal functions. In the early 1990s, we embarked on a research lobes, the number of individuals with focal frontal plan to examine whether such an executive system could pathology is not particularly high. The pressures of be fractionated (Stuss et al. 1995). Our approach was publication, the time required to collect an adequate different from that often used in neuropsychological sample of focal frontal lobe lesions and the rapid change research. Instead of selecting one test or one process that in theoretical positions during the period of data was considered executive, we took a ‘root and branches’ accumulation often predispose researchers to use more approach. We selected attention (the ‘root’) as the convenient samples (e.g. undifferentiated traumatic cognitive focus owing to its prominent role in many brain injury) as a proxy. Such studies have practical influential theories of frontal functions (e.g. Heilman & value for understanding the target population, but Watson 1977; Shallice 1982; Mesulam 1985; Norman & precise brain–behaviour relationships cannot be Shallice 1986; Posner & Petersen 1990; Knight 1991; determined. Central roles for various frontal regions Paus et al. 1997; Godefroy et al. 1999; Sturm & Willmes have been proposed for both cognitive and emotional 2001). We elected to study patients with focal lesions to demonstrate that a region was essential for an attentional * Author for correspondence ([email protected]). process, as opposed to simply being activated during One contribution of 14 to a Discussion Meeting Issue ‘Mental a process (as, say, with functional magnetic resonance processes in the human brain’. imaging, fMRI). All published studies addressing

901 This journal is q 2007 The Royal Society 762 Might there be clusters of control functions?Stuss Evidence from patient studies

22 Stuss (2011) J Int Neuropsyhcological Soc Fig. 1. This figure illustrates the frontal cortical—basal ganglia—thalamic circuits, supporting the fractionation of the frontal functional regions. Area 10 is not part of this circuitry, and is schematically presented in its polar location to suggest its integrative functions. For an expanded explanation of the anatomical and functional connections of Area 10 with other brain regions, see Gilbert, Gonen-Yaacovi, Benoit, Volle, & Burgess (2010) and Petrides and Pandya (2007). The figure also serves as a summary of the findings. STG 5 superior temporal gyrus; Right/Left 5 cerebral hemispheres.

There is other support for this model of discrete functional using similar operational definitions of processes as those out- categories within the frontal lobes. Comparative anatomical lined in the frontal patients. However, the demonstration of a studies and mapping of human brain development have parallel functional separation within the subcortical regions will identified two main frontal systems—a lateral one with pri- be difficult, because of the smaller size of these areas. For marily bidirectional connections to and from posterior cor- example, in several of our studies (Stuss et al., 1998, 2000), tices (executive) and an inferior/medial one with prominent patterns of performance after basal ganglia damage were similar limbic connections (emotional) (Pandya & Yeterian, 1996). to frontal patterns but, other than left–right differences, further These two systems are ‘‘energized’’ by the superior medial distinctions could not be isolated. One interesting approach region. The frontopolar region—both phylogenetically and has been the use of deep brain stimulation in the subthalamic ontogenetically late developing—integrates the executive nucleus to demonstrate an alteration in a frontotemporal and the emotional processes. Of the exquisitely mapped, network related to the performance of a verbal fluency task vertically segregated frontal–subcortical circuits (Alexander (Schroeder et al., 2003). et al., 1986), three align with our categories of energization, Similar questions could—and should—be raised about the executive, and emotional. The frontopolar region (integrative functional similarities and dissimilarities in other frontal function) does not have major frontal–subcortical connec- networks. We have pursued the question related to fronto- tions precisely because its role is integrating processes within cerebellar connectivity. If characterization of patients is strict, the frontal lobes and with other regions (Petrides & Pandya, and patients are studied in a chronic stage of recovery with 2007). See Figure 1. lesions limited to the cerebellum, the functional similarity is quite limited and specific (Alexander, Gillingham, Schwei- zer, & Stuss, in press; Schweizer, Alexander, Gillingham, Brain Systems and Networks Cusimano, & Stuss, 2010). The potential specific role of Our goal was to understand and fractionate the functions of white matter pathways also needs to be investigated. the frontal lobes. For each frontal cortical functional region, Understanding the role of specific brain regions within the there is a connection with a specific basal ganglia area, con- frontal lobes is not phrenology; analysis of the simple tasks tinuing to a defined-thalamic region. Are the functions of the and how different regions, frontal and non-frontal, are connected regions the same? This question needs to be pursued required depending on task demands and difficulty, identifies 60 MIYAKE ET AL.

Might there be clusters of control functions? Evidence from performance of healthy people across different tests of executive function 70 MIYAKE ET AL.

FIG. 1. (A) The theoretical ‘‘full, three-factor’’ model used for the confirmatory factor analysis (CFA). The ellipses represent the three executive functions (latent variables), and the rectangles represent the individual tasks (manifest variables) that were chosen to tap the spe- cific executive functions, as indicated by the straight, single-headed arrows. The curved dou- FIG. 2. The estimated three-factor model. Single-headed arrows have standardized factor ble-headed arrows represent correlations among the latent variables. Both models depict three loadings next to them. The loadings, all significant at the .05 level, are equivalent to standard- latent constructs, namely, Shifting, Updating, and Inhibition, which are hypothesized to be ized regression coefficients (beta weights) estimated with maximum likelihood estimation. correlated but separable. (B) A generic model for the structural equation modeling (SEM) The numbers at the ends of the smaller arrows are error terms. Squaring these terms gives analysis. This model is identical to the CFA model with the addition of a manifest variable an estimate of the variance for each task that is not accounted for by the latent construct. The on23 the right side that represents a complex executive function measure. In this particular model curved, double-headed arrows have correlation coefficients next to them and indicate signifi- (the ‘‘full’’ model), the manifest variable on the right has paths from all three latent variables cant correlations between the latent variables. Miyake et al (2000) Cognitive Psychology to estimate the contribution of each to performance on the executive task.

pendent factors’’ model, in which all the interfactor correlations are set to zero, should provide a fit to the data similar to that of the model in which the correlations are allowed to vary freely. The full three-factor model, complete with the estimated factor loadings, is illustrated in Fig. 2. The numbers next to the straight, single-headed arrows are the standardized factor loadings, and those next to the curved, double- headed arrows are the correlations between the factors. In addition, the num- bers at the ends of the smaller, single-headed arrows represent the error terms. Squaring these error terms gives an estimate of the unexplained vari- ance for each task, which could be attributed to idiosyncratic task demands and measurement error. Note that all the factor loadings listed in Fig. 2 are equivalent to standardized regression coefficients and can be interpreted ac- cordingly. The fit indices for this full three-factor model, summarized in Table 2 (Model 1), were all excellent. Specifically, this model produced a nonsig- nificant χ2(24, N ϭ 137) ϭ 20.29, p Ͼ .65, indicating that the model’s predic- Cognitive control is often term used in cognitive neuroscience One theory proposes that anterior cingulate cortex detects conflict and DLPFC exerts control

24 Macdonald et al (2000) Science NE39CH08-Hayden ARI 26 May 2016 13:46 Anterior cingulate monitors for conflict Deployed when input shows conflicting information – in non-default conditions

a Default action b Controlled action dACC dACC Monitoring Monitoring

Input Output Input Output

Controlled Controlled pathway pathway Stimuli Action Stimuli Action Default Default pathway pathway

25 c d Heilbronner & Hayden (2016) Ann Rev Neurosci Input Output Input Output

Other Other areas dACC areas dACC Stimuli Action Stimuli Action Other Other Other Other areas areas areas areas

Figure 3 Schematic illustrating two models of the role of the dorsal anterior cingulate cortex (dACC) in control. (a,b)Insomemodels,thedACC is seen as a controller and, thus, outside the standard input–output transformations that make up decision making, which are presumed to be housed in other brain areas. In such models, (a)default(i.e.,well-learnedoruncontrolled)actionsactivatethedACConlyweakly (indicated by thin lines)becausetheneedforcontrolisnotdetected,but(b) controlled actions activate the dACC strongly (indicated by thick lines) as the dACC monitors the need for control and summons it. (c,d ) In other models, the dACC serves as one part of the input–output transformation pathway, although it is not necessarily the only way for information to pass toward actions. In such models, (c) default actions activate the dACC weakly because input–output transformations are relatively efficient, but (d ) controlled actions activate the dACC more strongly because input–output transformations require more overall activation.

processes. Morecraft & Van Hoesen (1998), focusing on its anatomy, proposed that the dACC serves as an entry point for limbic information into the motor system. In contrast, Bush et al. (2000) argued that it serves to regulate cognitive and emotional processing. Aside from the inclusion of cognitive variables, the biggest difference among these models is that the Bush group saw the Annu. Rev. Neurosci. 2016.39:149-170. Downloaded from www.annualreviews.org dACC as a modulator of cognitive processes, whereas Morecraft & Van Hoesen saw it as an

Access provided by University of Oxford - Bodleian Library on 10/20/16. For personal use only. essential part of those processes, and a relatively late part because it directed motor action. Continuing these threads, Rushworth and colleagues (2011) have argued that the dACC serves to link actions with outcomes and, thus, to guide actions by offering motor cortex information about the consequences of possible actions. In contrast, Shenhav et al. (2013) have argued that the dACC integrates information relevant for control, and it signals to other regions how that control should be orchestrated (Figure 3a,b).

Contexts and Strategies The general view from the physiological literature is that individual dACC neurons track many task-related variables. We propose that these variables can be categorized into ones that reflect task state and ones that guide (or at least correlate with, if we are being cautious about inferring

www.annualreviews.org Dorsal Anterior Cingulate Cortex 161 •

CONEUR-1078; NO. OF PAGES 10

2 Decision making

Figure 1

(a) IOFC = value assignment REPORTS REPORTS

IOFC

+100 Major dispute about the function of ACC +70

Anterior cingulate cortex: adjusts behaviour to adapt to changing environments?

CONEUR-1078; NO. OF PAGES 10

ACC = mOFC =

(b)mOFC/vmPFC= (c) mOFC/vmPFC= § (d)Rushworth, Kolling and colleagues have proposed a key role for ACC in adjusting behaviour

2 Decision making value expectation search vs encounter specific value value comparison value comparison comparisons [based on IOFC information] [based on IOFC information] fromi.e. foraging persevering guidance i.e. exploitationwith current policy to adopting a new one. Analogy with animal foraging. Both compete to determine choices

Figure 1 Fig. 2. ACC activity was higher in forages than decisions (A), better related to mOFC/ ACC the inverse value difference (VD) during decisions than foraging (B), reflected (a) IOFC = value assignment vmPFC the main effect of search value during foraging (C), and related better to search VD than decision VD (D). ACC time courses during engage (E)andsearch(F). (G)IndividualpeakACCBOLDb weights 5 to 10 s after forage stimulus onset correlated with behavioral effects of the search value on search behavior

IOFC (bottom), whereas ACC b weights of best search value component predicted repeated searching (top). VmPFC exhibited no such correlations. (H)Timecourse +100 for engage forages and the subsequent decision phase: The search value (red) +70 signal continued into the decision phase. Reward magnitudes associated with Search chosen (green) and unchosen (orange) components of encounter value (left) were represented from their onset in the forage phase and into the decision phase. The Engage reward probabilities of the chosen and unchosen options were only revealed after ACC = mOFC = (b)mOFC/vmPFC= (c) mOFC/vmPFC= (d) Kolling et al (2012) Science ACC = value expectation search vs encounter specific value engaging, and their BOLD effects therefore appear later (right). valueACC= comparison value comparison comparisons [based on IOFC information] [based on IOFC information] value comparison action value comparison i.e. foraging guidance i.e. exploitation Fig. 2. ACC activity was higher in forages than decisions (A), better related to [based on mOFC information] Both compete to determine choices the inverse value difference (VD) during decisions than foraging (B), reflected ACC mOFC/ decisions were separated in time. When partic- vmPFC § According to this view: primary role of ACC is dynamically updatingFig. 3. behavioural(A)VmPFCtime policies in ipants chose to search, new options drawn at courses during forages the main effect of search value during foraging (C), and related better to search changing environments random from the boxed alternatives were en- (conventions as in Fig. 26 countered. Participants searched as often as they 2E). (B)Activitybetter- VD than decision VD (D). ACC time courses during engage (E)andsearch(F). Kolling et al (2016) Nat Neurosci wished but risked the same costs each time. related to decision VD Logistic regression identified factors weigh- than to forage VD. (C) (G)IndividualpeakACCBOLDb weights 5 to 10 s after forage stimulus onset Search ing on forages and decisions. Engaging was pro- VmPFC time course for- engage forages and the correlated with behavioral effects of the search value on search behavior Engage moted by search costs and encounter values but ? retarded by all components of search values (Fig. subsequent decision phase ACC = ACC= (bottom), whereas ACC b weights of best search value component predicted value comparison action value comparison (conventions as in Fig. 2H). [based on mOFC information] 1B and fig. S7). Participants were biased against search and required objectively more value gain (D)IndividualpeakvmPFC Current Opinion in Neurobiology BOLD b weights 5 to 10 s repeated searching (top). VmPFC exhibited no such correlations. (H)Timecourse for searching than engaging (the constant from

after decision onset cor-

the regression reflects subjects’ biases against for engage forages and the subsequent decision phase: The search value (red)

Values and reward type information are assigned to different options in lOFC (a). There are then different possible ways in which valuation and decision- related with estimates of

making mechanisms might proceed in frontal cortex. They might proceed in series with reward expectations represented in vmPFC/mOFC and the actual

searching; we call this parameter forage read- decision accuracy (soft-

process of value comparison occurring in ACC (b). Alternatively there may be distinct mechanisms for, first, making decisions about the rewards that iness). Decisions were influenced by reward proba- signal continued into the decision phase. Reward magnitudes associated with

should be the focus of behaviour and attention and, second, for making decisions about the actions that should be made to obtain those rewards (c). max temperature).

bility and magnitude differences between options

According to an alternative scheme there are parallel systems for foraging and decision-making (d). According to this last view the ACC, which exists albeit chosen (green) and unchosen (orange) components of encounter value (left) were

in different forms in all mammals, might guide foraging, a key behaviour for all mammals, while vmPFC, a primate specialization [49] might be especially (Fig. 1C).

important when decisions are made between simultaneously available options. Other authors have emphasized other differences between these regions,

? Comparison of average activity during forag-

represented from their onset in the forage phase and into the decision phase. The

for example, emphasizing that ACC may have a greater role in reward-guided learning because of the prominence of prediction errors in this brain region ing and decisions identified ACC among other

[5], the presence of reward outcome-related activity reflecting different mnemonic time constants [50], and surprise [51,52]. Current Opinion in Neurobiology regions (Fig. 2A). Usually, in decisions, the most reward probabilities of the chosen and unchosen options were only revealed after

common signal observed in ACC is inversely

Current Opinion inValuesNeurobiology and reward type2012,information22:1are–10assigned to different options in lOFC (a). There are then different possible ways in which valuation andwww.sciencedirect.comdecision-

making mechanisms might proceed in frontal cortex. They might proceed in series with reward expectations represented in vmPFC/mOFC and the actual related to the value difference between chosen engaging, and their BOLD effects therefore appear later (right).

process of value comparison occurring in ACC (b). Alternatively there may be distinct mechanisms for, first, making decisions about the rewards that

Please cite this article in press as: Rushworth MFS, et al.. Valuation and decision-making in frontal cortex: one or many serial or parallel systems?, Curr Opin Neurobiol (2012), http://dx.doi.org/

should be the focus of behaviour and attention and, second, for making decisions about the actions that should be made to obtain those rewards (c). and unchosen options. Such inverse value dif-

10.1016/j.conb.2012.04.011

According to an alternative scheme there are parallel systems for foraging and decision-making (d). According to this last view the ACC, which exists albeit ference effects have been interpreted as indicat-

in different forms in all mammals, might guide foraging, a key behaviour for all mammals, while vmPFC, a primate specialization [49] might be especially

important when decisions are made between simultaneously available options. Other authors have emphasized other differences between these regions, ing that ACC or dorsomedial frontal cortex is a

for example, emphasizing that ACC may have a greater role in reward-guided learning because of the prominence of prediction errors in this brain region

[5], the presence of reward outcome-related activity reflecting different mnemonic time constants [50], and surprise [51,52]. “comparator” comparing choice values. Accord-

ing to this theory, the region is more active when

Current Opinion in Neurobiology 2012, 22:1–10 www.sciencedirect.com unchosen values are larger, because a smaller

Please cite this article in press as: Rushworth MFS, et al.. Valuation and decision-making in frontal cortex: one or many serial or parallel systems?, Curr Opin Neurobiol (2012), http://dx.doi.org/ decisionsdifference were between separated chosen and unchosen in values time. When partic- Fig. 3. (A)VmPFCtime 10.1016/j.conb.2012.04.011 means comparison takes longer before a choice is benefit of the alternative course of action or the tary text1.5) and tested for a region that demon- ipantsmade chose (6, 7)(fig.S3).Relatedaccountsemphasize to search, new optionsvalue of exploring drawn the at environment.courses This hy- duringstrated both forages of these effects: Coding for the an ACC role in monitoring for conflict between pothesis predicts that ACC, during forages, will unchosen–chosen value difference during deci- randomresponses from (8). the boxed alternativesstop reflecting were the value en- of the unchosen(conventions option sions but as not forages in Fig. (Fig. 2B), and, on forages, However, our task also allowed us to test and will always represent the value2E). of searching. (B)Activitybetter-instead coding for the search value (Fig. 2C). countered.whether Participants the ACC signal reflects searched the relative asWe often therefore as refined they the analysis (supplemen- Both tests identified overlapping ACC regions. wished but risked the same costs each time. related to decision VD Logistic96 regression identified6APRIL2012 factors VOL336 weigh-SCIENCE thanwww.sciencemag.org to forage VD. (C) ing on forages and decisions. Engaging was pro- VmPFC time course for- moted by search costs and encounter values but engage forages and the retarded by all components of search values (Fig. subsequent decision phase 1B and fig. S7). Participants were biased against (conventions as in Fig. 2H). search and required objectively more value gain (D)IndividualpeakvmPFC BOLD b weights 5 to 10 s for searching than engaging (the constant from after decision onset cor- the regression reflects subjects’ biases against related with estimates of searching; we call this parameter forage read- decision accuracy (soft- iness). Decisions were influenced by reward proba- max temperature). bility and magnitude differences between options (Fig. 1C). Comparison of average activity during forag- ing and decisions identified ACC among other regions (Fig. 2A). Usually, in decisions, the most common signal observed in ACC is inversely related to the value difference between chosen and unchosen options. Such inverse value dif- ference effects have been interpreted as indicat- ing that ACC or dorsomedial frontal cortex is a “comparator” comparing choice values. Accord- ing to this theory, the region is more active when unchosen values are larger, because a smaller difference between chosen and unchosen values means comparison takes longer before a choice is benefit of the alternative course of action or the tary text1.5) and tested for a region that demon- made (6, 7)(fig.S3).Relatedaccountsemphasize value of exploring the environment. This hy- strated both of these effects: Coding for the an ACC role in monitoring for conflict between pothesis predicts that ACC, during forages, will unchosen–chosen value difference during deci- responses (8). stop reflecting the value of the unchosen option sions but not forages (Fig. 2B), and, on forages, However, our task also allowed us to test and will always represent the value of searching. instead coding for the search value (Fig. 2C). whether the ACC signal reflects the relative We therefore refined the analysis (supplemen- Both tests identified overlapping ACC regions.

96 6APRIL2012 VOL336 SCIENCE www.sciencemag.org Major dispute about the function of ACC PERSPECTIVE Anterior cingulate cortex: computes the expected value of control?

"" " We focus primarily on potential functional homologies within and $#" #" $#" adjacent to this circumscribed region of cortex, an approach that, ' "   " ! !# "  "& in spite of its benefits, risks blurring potential distinctions between "  ! ("  " "# species, methods and cytoarchitectonic boundaries. We will return "       #! ##! to these considerations later.

  !  *# ' dACC’s role in monitoring for EVC-relevant signals )" "  "  !"! "$ ! !+  """#! The EVC theory explains why dACC is more active in control-demanding 10,21 # !   !" "' situations ; for instance, those that require processes that are com- !""  plex, deliberate, novel, and/or exploratory versus habitual and/or exter- # ""' "$ !"$   "   ' %  ! !! nally driven. Overwhelming evidence links dACC activity to signals indicating a demand for control6,14,15,21—including errors22–26, explicit FigureAccording 1 dACC’s to Shenhav proposedand role Princeton in control colleagues, allocation based ACC detects on EVC. need Top: for controlnegative (as infeedback conflict27– monitoring)29, conflict 1,30–34 and surprise35,36—as well as its the EVC theory proposes that dACC monitors for information relevant to but also calculates its value and signal strength depends upon costs/benefitspayoff of13 control,37,38. Furthermore, dACC responses to these signals can carry evaluating EVC and specifies the optimal control allocation to downstream 27 information about the particular type of control that is required22,26,38 regions. dACC is shown in the central panel, based on a Neurosynth meta- Shenhav et al (2016) Nat Neurosci analysis of 428 human neuroimaging studies associated with cognitive and are weaker in situations that pose less of a demand for control (for control (http://neurosynth.org/analyses/terms/cognitive%20control/). example, when correct feedback is predicted28,29 or when a surprising Example input and output structures are shown at the left and right sides event does not bear on future task performance36). of the panel. Bottom: Previous findings suggest a role for dACC in using each of the monitoring signals listed (for example, errors, conflict) as the dACC’s role in specifying control based on EVC basis for one or more subsequent adjustments in control (for example, adjustment in one’s speed-accuracy tradeoff or attentional focus). The EVC theory extends performance-monitoring accounts of dACC As indicated in the central panel, the EVC theory proposes that dACC to explain how signals indicating the need for and/or value of control evaluates the expected future benefits of applying varying intensities of are translated into the adaptive execution of control (Fig. 1). In par- control (arrow in each gauge) for each candidate control signal (different ticular, the theory proposes that dACC uses estimates of EVC to select gauges; for example, different rules and the degree to which they could appropriate control signals, which are then implemented by other be attended) and subtracts from this the intrinsic cost of applying a neural structures to influence processing. This explains a number of given control intensity. This results in an estimate of the EVC. dACC then selects for execution the control signal settings that maximize EVC, findings tying performance-monitoring-related responses in dACC 4,14,15 projecting information about these signals to relevant downstream regions or nearby regions to subsequent control adjustments , including 25,39 responsible for implementing the corresponding signals. Elsewhere slower and/or more accurate responding ; increased allocation of we have provided a formal description of this process that aligns EVC attention toward task-relevant stimulus properties and away from and its components with ideas from artificial intelligence and control irrelevant ones14,31,33,34; and adaptive changes in switching behav- 4,74 theory and shown how a computational implementation of the theory ior39, response bias40 and the pace of learning35,41 (Fig. 1). can account for observed influences of incentives on control adjustments The proposition that dACC is responsible for specifying control (and associated behavior)74. OFC: orbitofrontal cortex; STN: subthalamic nucleus; mPFC: medial prefrontal cortex; PFC: prefrontal cortex. Center signals required to execute control-demanding tasks also explains cor- panel adapted with permission from ref. 4, Elsevier. relations observed between task performance and dACC’s encoding of the task environment5,42. For instance, the strength of rule encoding in monkey dACC predicts whether that rule will be executed properly implementation of control10. The EVC theory identifies and makes on a forthcoming trial43, appears earlier on trials following an error43 computationally explicit an interface that is common across most of and weakens with more repetitions following a rule switch (as lateral these theoretical perspectives, between the monitoring for control- prefrontal cortex rule selectivity increases)44. Similarly, in humans, relevant information and the allocation of control. In this respect, it stronger white matter tracts between dACC and other regions are can be viewed as an extension of its historical predecessor, the conflict associated with improved cognitive control45,46. monitoring theory1. This theory proposed that dACC monitors ongo- This proposition also explains why causal manipulations involving ing processing, including the amount of conflict between potential dACC can influence one’s ability to adjust control. Inactivating or responses, for signals indicative of the need for control. However, lesioning dACC can impair within-trial error correction47 and post- it made highly simplified assumptions about how those signals are error slowing25, decrease conflict-related adaptation31,48 and impair evaluated and used to allocate control. adaptive reversal between responses49,50 and task sets51. Moreover, The EVC theory extends this monitoring framework to address how dACC inactivation impairs antisaccade performance52 whereas this process occurs in greater detail (Fig. 1). In doing so, it provides microstimulation facilitates it53. Similar effects are found when tran- a more comprehensive account of dACC function, accommodating scranial direct current stimulation (tDCS) is used to modulate activity previously unaddressed empirical data (for example, dACC’s asso- within dACC and surrounding cortex54,55. Inhibitory tDCS produces ciations with reward and motivation) and filling explanatory gaps weaker error- and feedback-related dACC responses and concomitant regarding control allocation (for example, how the demanded control impairments in performance (accuracy, learning rate and post-error is deemed worth allocating). Because much of the force of the theory slowing); excitatory tDCS produces the opposite patterns54. While the lies in its ability to integrate a wide range of findings that have previ- tDCS configuration in these studies likely resulted in spillover modu- ously been considered to reflect disparate functions, we provide a lation of other regions (including parts of lateral prefrontal cortex), brief review of these findings below (see Fig. 1 and ref. 4 for a fuller preventing strong inference about the locus of influence, these find- discussion). While necessarily selective, this review of the literature ings point to a potentially valuable avenue for noninvasively studying attempts to highlight convergent findings across species and meas- causal relationships between human dACC and control allocation. urement techniques within regions potentially homologous to those This would supplement studies of lesion patients, which can produce highlighted in Figure 1, in particular anterior midcingulate cortex. variable findings (for example, ref. 56) partly due to factors out of the

NATURE NEUROSCIENCE VOLUME 19 | NUMBER 10 | OCTOBER 2016 1287 NE39CH08-Hayden ARI 26 May 2016 13:46

a Default action b Controlled action dACC dACC Monitoring Monitoring NE39CH08-Hayden ARI 26 May 2016 13:46 Input Output AnteriorInput cingulate theoriesOutput Controlled Also put in perspective ofControlled offline controller versus online transformation of inputs to outputs pathway pathway Stimuli Action Stimuli Action Default Foraginga valueDefault theoryDefault action Expectedb valueControlled of control action pathway dACCpathway dACC Monitoring Monitoring c d Input Output InputInput Output Output Input Output

Other Other dACC Controlled dACC Controlled areas areas pathway pathway Stimuli Action StimuliStimuli Action Action Stimuli Action Other Other Other Default Other Default areas areas areas pathway areas pathway

Figure 3 c d Schematic illustrating two models of the role of the dorsal anterior28 cingulate cortex (dACC) in control. (a,b)Insomemodels,thedACC Input Output Input Output is seen as a controller and, thus, outside the standard input–output transformations that make up decision making, which are presumed Heilbronner & Hayden (2016) Ann Rev Neurosci to be housed in other brain areas. In such models, (a)default(i.e.,well-learnedoruncontrolled)actionsactivatethedACConlyweaklyOther Other (indicated by thin lines)becausetheneedforcontrolisnotdetected,but(b) controlledareas actions activate thedACC dACC strongly (indicated by areas dACC thick lines) as the dACC monitors the need for control and summons it.Stimuli (c,d ) In other models, the dACC serves as one partAction of the Stimuli Action input–output transformation pathway, although it is not necessarily the only way for information to pass toward actions. In such models, (c) default actions activate the dACC weakly because input–output transformationsOther are relativelyOther efficient, but (d ) controlled Other Other actions activate the dACC more strongly because input–output transformations requireareas more overall activation.areas areas areas processes. Morecraft & Van Hoesen (1998), focusing onFigure its anatomy, 3 proposed that the dACC serves as an entry point for limbic information into the motorSchematic system. illustrating In contrast, two Bush models et al. of the (2000) role of the dorsal anterior cingulate cortex (dACC) in control. (a,b)Insomemodels,thedACC is seen as a controller and, thus, outside the standard input–output transformations that make up decision making, which are presumed argued that it serves to regulate cognitive and emotional processing.to be housed inAside other from brain the areas. inclusion In such models, of (a)default(i.e.,well-learnedoruncontrolled)actionsactivatethedACConlyweakly cognitive variables, the biggest difference among these models is that the Bush group saw the

Annu. Rev. Neurosci. 2016.39:149-170. Downloaded from www.annualreviews.org (indicated by thin lines)becausetheneedforcontrolisnotdetected,but(b) controlled actions activate the dACC strongly (indicated by dACC as a modulator of cognitive processes, whereas Morecraftthick lines) as & the Van dACC Hoesen monitors saw the it need as an for control and summons it. (c,d ) In other models, the dACC serves as one part of the input–output transformation pathway, although it is not necessarily the only way for information to pass toward actions. In such Access provided by University of Oxford - Bodleian Library on 10/20/16. For personal use only. essential part of those processes, and a relatively late part because it directed motor action. models, ( ) default actions activate the dACC weakly because input–output transformations are relatively efficient, but ( ) controlled Continuing these threads, Rushworth and colleagues (2011) havec argued that the dACC serves d actions activate the dACC more strongly because input–output transformations require more overall activation. to link actions with outcomes and, thus, to guide actions by offering motor cortex information about the consequences of possible actions. In contrast, Shenhav et al. (2013) have argued that the dACC integrates information relevant for control, and it signalsprocesses. to other Morecraft regions & how Van that Hoesen control (1998), focusing on its anatomy, proposed that the dACC should be orchestrated (Figure 3a,b). serves as an entry point for limbic information into the motor system. In contrast, Bush et al. (2000) argued that it serves to regulate cognitive and emotional processing. Aside from the inclusion of cognitive variables, the biggest difference among these models is that the Bush group saw the Contexts and Strategies Annu. Rev. Neurosci. 2016.39:149-170. Downloaded from www.annualreviews.org dACC as a modulator of cognitive processes, whereas Morecraft & Van Hoesen saw it as an The general view from the physiological literature is that individual dACC neurons track many

Access provided by University of Oxford - Bodleian Library on 10/20/16. For personal use only. essential part of those processes, and a relatively late part because it directed motor action. task-related variables. We propose that these variables can beContinuing categorized these into threads, ones that Rushworth reflect and colleagues (2011) have argued that the dACC serves task state and ones that guide (or at least correlate with, ifto we link are actions being cautiouswith outcomes about inferring and, thus, to guide actions by offering motor cortex information about the consequences of possible actions. In contrast, Shenhav et al. (2013) have argued that the www.annualreviews.org Dorsal Anterior Cingulate Cortex 161 dACC integrates• information relevant for control, and it signals to other regions how that control should be orchestrated (Figure 3a,b).

Contexts and Strategies The general view from the physiological literature is that individual dACC neurons track many task-related variables. We propose that these variables can be categorized into ones that reflect task state and ones that guide (or at least correlate with, if we are being cautious about inferring

www.annualreviews.org Dorsal Anterior Cingulate Cortex 161 • Neuron

Neuron Review ReviewMultiple demand network Executive functions aren’t necessarily ‘frontal’ – a fronto-parietal system identified across studies

A B A B premotor cortex Hard vs easy contrastspremotor cortex intraparietal pre-SMA / anterior cingulate intraparietal pre-SMA / anterior cingulate sulcus sulcus Remembering words / non-words inferior frontal inferior frontal sulcus sulcus

Arithmetic

Spatial working memory anterior insula / frontal operculum

anterior insula / VerbalC working memory frontal operculum

C

29 Three different ‘conflictMD tasks’ language Duncan (2013) Neuron Figure 2. Multiple-Demand System in the Human Brain (A) Activity pattern for hard minus easy contrast in tasks tapping multiple cognitive domains (top to bottom: remembering word/nonword strings, arithmetic, spatial working memory, verbal working memory, and three versions of resisting response conflict). (B) Mean hard minus easy activity pattern across all seven tasks. Generally bilateral activity has been captured by averaging across left and right hemispheres, and projecting the resulting mean image onto the right. Included in the full multiple-demand (MD) pattern are a posterior strip of the lateral frontal surface, from premotor cortex to the frontal operculum and anterior insula; an anterior-posterior strip extending along the inferior frontal sulcus; a dorsomedial strip extending from pre-SMA to the dorsal part of the anterior cingulate; and activity along the length of the intraparietal sulcus. At least in visual tasks, accompanying activity is generally seen in higher visual areas. Outside the cerebral cortex, activity is also seen in the medial cerebellum and elsewhere. (C) Left hemisphere activity for two example participants, showing closely adjacent MD (blue; greater activity in memory for nonword stringsMD versus sentences) language and language (red; reverse contrast) regions. Adapted with permission from Fedorenko et al. (2013).

Figure 2. Multiple-Demand System in the Human Brain More telling are the(A) resultsActivity of pattern single-subject for hard analyses. minus Using easy contrastTo link in imaging tasks tappingresults to multipleelectrophysiology, cognitive a criticaldomains question (top to bottom: remembering word/nonword strings, arithmetic, one task as a localizerspatial for MD working voxels memory, in a single verbal subject’s working brain, memory,is correspondence and three between versions human of resisting and animal response systems. conflict). In the Fedorenko et al. (2013)(B) Meanwenton hard to askminus how easy these activity same voxels pattern acrossmacaque, all oneseven fMRI tasks. study Generally comparing bilateral pro- and activity antisaccades has been captured by averaging across left and right hemispheres, behaved in the otherand six projecting tasks (see the also resultingStiers meanet al., image 2010; ontoshowed the right.activity Included in lateral frontal, in the full dorsomedial multiple-demand frontal, and (MD) parie- pattern are a posterior strip of the lateral frontal surface, from Wojciulik and Kanwisher,premotor 1999 cortex). For all to major the frontal regions operculum shown in andtal cortex,anterior reminiscent insula; an ofanterior-posterior the human MD pattern strip extending (Ford et al., along the inferior frontal sulcus; a dorsomedial strip extending Figure 2B, resultsfrom were pre-SMA similar; in to individual the dorsal subjects, part of voxels the anterior2009 cingulate;). Somewhat and similar activity patterns along have the alsolength been of the obtained intraparietal by sulcus. At least in visual tasks, accompanying activity is with strong responsegenerally to the seen localizer in higher also showed visual increasedareas. Outsidea simple the cerebral comparison cortex, of activity visual stimulation is also seen versus in the fixation medial cerebellum and elsewhere. activity for other cognitive(C) Left demands. hemisphere In individual activity subjects, for two example tell- (Stoewer participants, et al., 2010 showing) and by closely analysis adjacent of correlations MD (blue; in resting greater activity in memory for nonword strings versus sentences) ingly, MD regions wereand language often immediately (red; reverse adjacent contrast) to regions regions.state Adapted data (Sallet with et permission al., 2013; though from seeFedorenkoMantini et et al., al. 2013 (2013)). . with a very different functional profile, responding to increased In support of these functional data, anatomical studies confirm difficulty neither in the localizer nor in other tasks. In left lateral connections between lateral frontal, dorsomedial frontal, frontal cortex, for example, a typical MD region often surrounded parietal, and insular regions (e.g., Cavada and Goldman-Rakic, a quite different region, showing selective activity for language 1989; Goldman-Rakic, 1988; Mufson and Mesulam, 1982). (Fedorenko et al., 2012More; see Figure telling 2C). are In a the control results region of of single-subjectActivity crossing analyses. multiple task Using demands suggestsTo link functions imaging of results to electrophysiology, a critical question the temporal lobe, equivalentone task analyses as a localizer suggested no for MD MD voxels voxelsimportance in a single in many subject’s different kinds brain, of cognition.is correspondence In the following between human and animal systems. In the in single subjects, i.e., no voxels with common response to mul- sections, I suggest that the core function of the MD system is to tiple cognitive demands.Fedorenko Taken together, et al. these (2013) data showwent tightly on tocontrol ask how complex these behavior same in a voxelsseries of attentionalmacaque, episodes. one In fMRI study comparing pro- and antisaccades localized MD activity,behaved varying in exact in the pattern other from one six person tasks to (seemultiple also MD regions,Stiers I suggestet al., that 2010; neurons haveshowed highly dynamic activity in lateral frontal, dorsomedial frontal, and parie- another but with a highlyWojciulik consistent and overall Kanwisher, topography 1999in frontal). Forresponse all major properties, regions adapting shown to code in the specifictal cortex, information reminiscent of the human MD pattern (Ford et al., and parietal cortex. and events within the current attentional focus (Duncan, 2001). Figure 2B, results were similar; in individual subjects, voxels 2009). Somewhat similar patterns have also been obtained by with strong response to the localizer also showed increased a simple comparison of visual stimulation versus fixation Neuron 80, October 2, 2013 ª2013 Elsevier Inc. 37 activity for other cognitive demands. In individual subjects, tell- (Stoewer et al., 2010) and by analysis of correlations in resting ingly, MD regions were often immediately adjacent to regions state data (Sallet et al., 2013; though see Mantini et al., 2013). with a very different functional profile, responding to increased In support of these functional data, anatomical studies confirm difficulty neither in the localizer nor in other tasks. In left lateral connections between lateral frontal, dorsomedial frontal, frontal cortex, for example, a typical MD region often surrounded parietal, and insular regions (e.g., Cavada and Goldman-Rakic, a quite different region, showing selective activity for language 1989; Goldman-Rakic, 1988; Mufson and Mesulam, 1982). (Fedorenko et al., 2012; see Figure 2C). In a control region of Activity crossing multiple task demands suggests functions of the temporal lobe, equivalent analyses suggested no MD voxels importance in many different kinds of cognition. In the following in single subjects, i.e., no voxels with common response to mul- sections, I suggest that the core function of the MD system is to tiple cognitive demands. Taken together, these data show tightly control complex behavior in a series of attentional episodes. In localized MD activity, varying in exact pattern from one person to multiple MD regions, I suggest that neurons have highly dynamic another but with a highly consistent overall topography in frontal response properties, adapting to code the specific information and parietal cortex. and events within the current attentional focus (Duncan, 2001).

Neuron 80, October 2, 2013 ª2013 Elsevier Inc. 37 Neuron IntelligenceReview and executive function Strong relationship between ‘fluid intelligence’ and multiple demand network?

A B premotor cortex intraparietal pre-SMA / anterior cingulate sulcus inferior frontal sulcus

anterior insula / frontal operculum Fluid intelliegence tasks Duncan TICS 2010 C

30 Duncan (2013) Neuron

MD language

Figure 2. Multiple-Demand System in the Human Brain (A) Activity pattern for hard minus easy contrast in tasks tapping multiple cognitive domains (top to bottom: remembering word/nonword strings, arithmetic, spatial working memory, verbal working memory, and three versions of resisting response conflict). (B) Mean hard minus easy activity pattern across all seven tasks. Generally bilateral activity has been captured by averaging across left and right hemispheres, and projecting the resulting mean image onto the right. Included in the full multiple-demand (MD) pattern are a posterior strip of the lateral frontal surface, from premotor cortex to the frontal operculum and anterior insula; an anterior-posterior strip extending along the inferior frontal sulcus; a dorsomedial strip extending from pre-SMA to the dorsal part of the anterior cingulate; and activity along the length of the intraparietal sulcus. At least in visual tasks, accompanying activity is generally seen in higher visual areas. Outside the cerebral cortex, activity is also seen in the medial cerebellum and elsewhere. (C) Left hemisphere activity for two example participants, showing closely adjacent MD (blue; greater activity in memory for nonword strings versus sentences) and language (red; reverse contrast) regions. Adapted with permission from Fedorenko et al. (2013).

More telling are the results of single-subject analyses. Using To link imaging results to electrophysiology, a critical question one task as a localizer for MD voxels in a single subject’s brain, is correspondence between human and animal systems. In the Fedorenko et al. (2013) went on to ask how these same voxels macaque, one fMRI study comparing pro- and antisaccades behaved in the other six tasks (see also Stiers et al., 2010; showed activity in lateral frontal, dorsomedial frontal, and parie- Wojciulik and Kanwisher, 1999). For all major regions shown in tal cortex, reminiscent of the human MD pattern (Ford et al., Figure 2B, results were similar; in individual subjects, voxels 2009). Somewhat similar patterns have also been obtained by with strong response to the localizer also showed increased a simple comparison of visual stimulation versus fixation activity for other cognitive demands. In individual subjects, tell- (Stoewer et al., 2010) and by analysis of correlations in resting ingly, MD regions were often immediately adjacent to regions state data (Sallet et al., 2013; though see Mantini et al., 2013). with a very different functional profile, responding to increased In support of these functional data, anatomical studies confirm difficulty neither in the localizer nor in other tasks. In left lateral connections between lateral frontal, dorsomedial frontal, frontal cortex, for example, a typical MD region often surrounded parietal, and insular regions (e.g., Cavada and Goldman-Rakic, a quite different region, showing selective activity for language 1989; Goldman-Rakic, 1988; Mufson and Mesulam, 1982). (Fedorenko et al., 2012; see Figure 2C). In a control region of Activity crossing multiple task demands suggests functions of the temporal lobe, equivalent analyses suggested no MD voxels importance in many different kinds of cognition. In the following in single subjects, i.e., no voxels with common response to mul- sections, I suggest that the core function of the MD system is to tiple cognitive demands. Taken together, these data show tightly control complex behavior in a series of attentional episodes. In localized MD activity, varying in exact pattern from one person to multiple MD regions, I suggest that neurons have highly dynamic another but with a highly consistent overall topography in frontal response properties, adapting to code the specific information and parietal cortex. and events within the current attentional focus (Duncan, 2001).

Neuron 80, October 2, 2013 ª2013 Elsevier Inc. 37 238 | Brain 2010: 133; 234–247 M. Roca et al. 238 | Brain 2010: 133; 234–247 M. Roca et al.

range = 19–70) and mean National Adult Reading Test-estimated IQ each patient, we calculated the percentage of each region included range = 19–70) and mean National Adult Readingwas Test-estimated 109.6 (SD = 12.3). IQ each patient, we calculated the percentage of eachin the region lesion. included For subgroup analysis, patients were assigned to one of was 109.6 (SD = 12.3). in the lesion. For subgroup analysis, patients werethe assigned four subgroups to one of based on which region had the greatest percent- Neuroradiological assessmentthe four subgroups based on which region had theage greatest damage. percent- Using this criterion, seven patients were classified as age damage. Using this criterion, seven patientsinferior were medial, classified eight as as superior medial, seven as left lateral and 14 Neuroradiological assessment MRI scans were performed for all patients and interpreted by a inferior medial, eight as superior medial, seven asas left right lateral lateral and (Table 14 1). Patients with lesions involving more than MRI scans were performed for all patients andneurologist interpreted with by experience a in structural neuroimaging, who was as right lateral (Table 1). Patients with lesions involving20% of each more of than two or more regions were excluded from subgroup neurologist with experience in structural neuroimaging,blind to the who experimental was results (FM). Lesions were traced using 20% of each of two or more regions were excludedcomparisons from subgroup (n = 8; Table 1, ‘multiple’). Lesion overlaps for left lat- blind to the experimental results (FM). LesionsMRIcro were traced (Rorden using and Brett, 2000; http://www.sph.sc.edu/comd/ comparisons (n = 8; Table 1, ‘multiple’). Lesion overlapseral, right for lateral, left lat- inferior medial and superior medial subgroups are MRIcro (Rorden and Brett, 2000; http://www.sph.sc.edu/comd/rorden/mricro.html) and normalized to a standard template using sta- eral, right lateral, inferior medial and superior medialshown subgroups in Fig. 2. are rorden/mricro.html) and normalized to a standardtistical template parametric using sta- mapping-5 software (Wellcome Department of shown in Fig. 2. tistical parametric mapping-5 software (WellcomeImaging Department Neuroscience, of London, England; www.fil.ion.ucl.ac.uk) with Imaging Neuroscience, London, England; www.fil.ion.ucl.ac.uk)cost-function masking with to mask the lesion from the calculation of the Neuropsychological assessment cost-function masking to mask the lesion from thenormalization calculation of parameters the Neuropsychological (Brett et al., 2001). Using assessment the Brodmann normalization parameters (Brett et al., 2001). Usingarea themaps Brodmann provided with MRIcroN (http://www.sph.sc.edu/comd/ Culture Fair (Institute for Personality and Ability Testing, Culture Fair (Institute for Personality and Ability Testing, area maps provided with MRIcroN (http://www.sph.sc.edu/comd/rorden/mricron), templates were created for four frontal regions: 1973) rorden/mricron), templates were created for fourinferior frontal medial regions: (Brodmann1973) area 24, 25, 32, plus medial parts of To assess fluid intelligence we used the Culture Fair, Scale 2 Form A, a inferior medial (Brodmann area 24, 25, 32, plusBrodmann medial area parts 10, of 11, allTo extending assess fluid up intelligence to level of we genu used of the corpus Culture Fair,standard Scale 2 test Form of A, novel a problem solving with a loading of 0.81 on a Brodmann area 10, 11, all extending up to levelcallosum); of genu ofsuperior corpus medialstandard (Brodmann test area of novel 24, 25, problem 32 and solving medial with10, a loadinggeneral of intelligence 0.81 on a factor (Institute for Personality and Ability Testing, callosum); superior medial (Brodmann area 24, 25,11, 32 all and above medial level 10, of genu,general plus medial intelligence parts factor of Brodmann (Institute area for 6,Personality 8, 1973). and Ability The test Testing, has four timed sets of problems (series completions, 11, all above level of genu, plus medial parts of Brodmann9); left lateral area (Brodmann 6, 8, 1973). area 43–47, The test plus has lateral four parts timed of sets Brodmann of problemsodd-one-out, (series completions, matrices, topological relations), all involving geometri- IQ and executive function in frontal9); left lateral lesion (Brodmann area 43–47, pluscases lateralarea parts 6, of 8, Brodmann 9, 10, 11); andodd-one-out, right lateral matrices, (Brodmann topological area as relations), for left allcal involving figures. geometri- Scores were converted to IQs using the area 6, 8, 9, 10, 11); and right lateral (Brodmannlateral). area To as separate for left lateralcal from figures. medial we Scores used a fixedwereX coordinate converted tostandardized IQs using table the of norms (Institute for Personality and Ability lateral). To separate lateral from medial we used aof fixed15X incoordinate Montreal Neurologicalstandardized Institute table (MNI) of norms atlas (Institute space. For for PersonalityTesting, 1973). and Ability Performance on some tests explained entirely by Culture Fairof 15 IQ in Montreal Neurological Institute (MNI)Æ atlas space. For Testing, 1973). Æ

Fluid intelligence in frontal lesions Brain 2010: 133; 234–247 | 241

Figure 3 Experiment 1. Regressions of Wisconsin Card Sorting Test and verbal fluency on Culture Fair IQ. Points show data for single patients (coloured) and controls (empty); regression line is calculated on combined patient and control data.

t(75) = 3.48, P50.001; Wisconsin Card Sorting Test, t(72) = 2.27, Verbal Fluency score positive for all four subtests (median 0.43), sug-

P50.02; and Verbal Fluency, t(74) = 3.09, P50.005. gesting behaviour similar to that of total IQ scores from the fullFigure test. 2 Lesion overlaps for patients with predominantly inferior medial, superior medial, left lateral and right lateral lesions. Colour Also as expected,31 both Wisconsin Card Sorting Test and Verbal Figure 2 Lesion overlaps for patients with predominantlyscales show inferior numbers medial, of affected superior patients medial, for left each lateral brain and voxel. right lateral lesions. Colour scales show numbers of affected patients for each brain voxel. Fluency were correlated with Culture Fair. Combining data from Experiment 2 Roca et al (2010) Brain patients and controls, Pearson’s correlations and (one-tailed) significance levels were r = 0.61, P50.001 for Wisconsin Card Again, one-tailed t-tests were used to compare patients (n = 21) À Sorting Test, and r = 0.56, P50.001 for Verbal Fluency and controls (n = 25). Results are shown in Table 2. The frontal (Table 2). Scatterplots are shown in Fig. 3, showing that higher group was significantly impaired on three subtests of the Ineco Culture Fair IQ was strongly associated with better performance in Frontal Screening: Go–No go, t(44) = 2.54, P50.01; Proverbs, both executive tasks. t(38) = 3.88, P50.001; Hayling, t(38) = 2.64, P50.01. One other The scatterplots suggest that, for these two executive tasks, (Spatial Working Memory) was marginal, t(38) = 1.67, P50.06. frontal deficits were entirely explained by fluid intelligence Significant differences were also found for deviation from optimal (cf. Fig. 1A). The effect of the frontal lesion was simply to shift time allocation on the Hotel Task t(43) = 3.82, P50.001, in the the Culture Fair distribution downward, without changing its Faux Pas t(43) = 2.95, P50.005 and in the Iowa Gambling Task relation to executive task performance. To assess this conclusion, t(41) = 1.69, P50.05. t-tests comparing patients and controls were repeated following For all tasks, correlations with Culture Fair were positive, adjustment for Culture Fair IQ as a covariate (equivalent to ana- showing better performance associated with higher IQ (Table 2). lysis of covariance with two-level factor patients versus controls). The correlation was significant for 9/12 tasks, including the six For both tasks, the difference between patients and controls was with significant difference between patients and controls. no longer significant; for the Wisconsin Card Sorting Test, For these six tasks, scatterplots relating performance to Culture t(71) = 0.35, P = 0.36, and for Verbal Fluency, t(73) = 1.48, Fair score are shown in Fig. 4. Contrary to the results from P = 0.07, both tests again one-tailed (Table 2). Experiment 1, these scatterplots suggest some difference between Figure 3 also suggests little difference between left lateral, right patients and controls even when correcting for the difference in lateral, inferior medial and superior medial subgroups. In confir- IQ (cf. Fig. 1B). As before, additional t-tests compared patients mation, ANOVA showed no significant difference between these and controls after adjusting for IQ scores. For the Iowa Gambling groups, for Culture Fair IQ, F(3, 32) = 1.76, P = 0.17, for Wisconsin Task, adjustment removed the significant patient–control differ- Card Sorting Test, F(3, 30) = 0.16, P = 0.92, or for Verbal Fluency, ence, t(40) = 1.07, P = 0.29. For the remaining five tasks, however, F(3, 31) = 0.76, P = 0.53 (Table 2, rightmost column; ANOVAs on significant differences remained even after such adjustment raw scores unadjusted for Culture Fair). (Table 2). As frontal lesions are sometimes specifically linked with Again, scatterplots in Fig. 4 suggest no evident differences perseverative errors on the Wisconsin Card Sorting Test (Milner, between left lateral, right lateral, inferior medial and superior 1963), analyses were repeated using percentage perseverative medial subgroups. Though the result is tempered by small subject errors (Nelson, 1976) instead of the total error score. For this numbers, in particular for inferior medial and superior medial, measure, the group difference between patients and controls ANOVA showed no significant difference between subgroups in was not significant, t(72) = 0.17, P = 0.43, accompanied by a any of the 12 tasks of Experiment 2 (Table 2). Given previous weaker correlation with Culture Fair, r = 0.31, P50.005. Again, suggestions that inferior medial damage may be especially impor- À ANOVA showed no significant difference between frontal tant in social or emotional functions, supplementary tests com- subgroups, F(3, 30) = 0.28, P = 0.84. pared all patients with any inferior medial damage (n = 8) to In a further subsidiary analysis, numbers of problems correctly remaining patients (n = 12), separately for the Iowa Gambling solved were examined for each of the four separate subtests of the Task, Mind in the Eyes and Faux Pas. In no case was the differ- Culture Fair. The correlation with Wisconsin Card Sorting Test total ence close to significant; for the Iowa Gambling Task errors was negative for all four subtests (median 0.52), and with t(16) = 0.81, P = 0.79 (data unavailable for two patients without À À 242 | Brain 2010: 133; 234–247 M. Roca et al.

242 | Brain 2010: 133; 234–247 M. Roca et al.

IQ and executive function in frontal lesion cases But performance on other tests not fully accounted for by Culture Fair IQ Fluid intelligence in frontal lesions Brain 2010: 133; 234–247 | 243

Fluid intelligence in frontal lesions Brain 2010: 133; 234–247 | 243

Figure 5 Experiment 2. Lesion overlap for 6 patients with worst average residual (performance adjusting for fluid intelligence) across Go–no go, Proverbs, Hayling, Hotel and Faux Pas tests. Left: overlap projected to brain surface; colour scale shows number of affected patients. Right: slice illustrating maximum overlap; coordinates in MNI space. Figure 4 Experiment 2. Regressions on fluid intelligence for all tasks showing significant difference between patients and controls. Symbols and regressions as Fig. 3. were significant for Go–No go, Proverbs, Hayling, and Spatial frontal lesions (e.g. Baldo and Shimamura, 1998; Davidson Working Memory; correlations with Culture Fair were significant et al., 2007), though are commonly stronger on the left (e.g. for Go–No go, Digit Span, Months, and Hayling, P 0.06 for Perret, 1974; Baldo and Shimamura, 1998). Further work would inferior medial damage), Mind in the Eyes t(18) = 0.52, P = 0.70, suggestion of selective5 association with anterior frontal lesions, Proverbs;À and no subtest showed significant differences between be needed to show why, in our patient sample, there was no and Faux Pas t(18) = 1.00, P = 0.33. left lateral, right lateral,especially inferior medial in the and right superior hemisphere. medial specific association with left hemisphere lesions. One possibility In a further analysis we employed a lesion overlapsubgroups. method for To examine these results in the wholeis that, patient by comparison group, for with each many previous studies, our lesion more targeted examination of deficits beyond those explained by patient we measured volume of damagesample was in four relatively segments anterior, of with no lesions extending into the Figure 5 Experiment 2. Lesion overlap for 6 patients with worst average residual (performance adjusting for fluid intelligence) across Lesion overlap for 6 patients with worst average residual (performance adjusting temporal lobe and lesions in only two of our left lateral patients Figure 4 Experiment 2. Regressions on fluid intelligence for allGo–no tasks showing go, Proverbs, significant Hayling,fluid difference intelligence. Hotel between and Faux First, patients Pas wetests. examined and Left: controls. overlap the relationship projected to brainbetween surface; colourthe frontal scale shows lobe, anterior number ofand affected posterior to Y = 35 in left and right incorporating more than 10% of Brodmann area 44. Previously, it Symbols and32 regressionsfor fluid as intelligence) Fig. 3. across Go–patients.no go, Right: Proverbs, slice illustratingsuch Hayling deficits maximum in, Hotel the overlap; five and tasks coordinates Faux where Pas they in MNI tests. were space.Discussion found (Go–No hemispheres. Mean volume of damage, and variability across has been suggested that frontal deficits in Verbal Fluency may be go, Proverbs, Hayling, Hotel, Faux Pas). For each task, scores patients, were similar for left anterior, right anterior and left In our data, the picture of relationsRoca between et al executive (2010) Brain deficit and over and above those predicted by general intelligence as mea- for each patient and control were convertedfluid to residuals intelligence after is both simpleposterior and unanticipated. regions, but Certainly, somewhat the greatersured by for the right Wechsler posterior. Adult Intelligence Scale (Henry and inferior medial damage), Mind in the Eyes t(18) = 0.52, P = 0.70,were significantsuggestion for of selective Go–Noadjusting go, association for Proverbs, fluid with intelligence Hayling, anterior and frontal (Figs Spatial 3lesions, and 4,frontalresults vertical lesions show distances that (e.g. fluid Baldo intelligenceFor and each is Shimamura, a region, substantial volumes contributor 1998; of Davidson damage to Crawford, were correlated 2004). Our with results mean suggest that fluid intelligence, with À and Faux Pas t(18) = 1.00, P = 0.33. Workingespecially Memory; in the correlations rightfrom hemisphere. the with regression CultureFair line). were Residuals significant in the fiveetfrontal al tasks., 2007), deficits. were though then For one are groupperformance commonly of ‘executive’ stronger residuals tasks, on including (deficits the left the beyond (e.g.its emphasis the prediction on current from problem fluid solving, may be the more suitable In a further analysis we employed a lesion overlap method forfor Go–NoTo examine go, Digit these Span,correlated, results Months, in for the and a whole total Hayling, patient of 10P group,5 correlations0.06 for for each betweenPerret,Wisconsin 1974; all Card possible Baldo Sorting and Test, Shimamura,intelligence, Verbal Fluency 1998). averaged and Further Iowa across Gambling, work the would fivemeasure critical of g tasks)in frontal as patients above. (Duncan et al., 1995). differences between patients and controls can be entirely g more targeted examination of deficits beyond those explained byProverbs;patient and we no subtest measuredtask showed pairsvolume significant (see of damageSupplementary differences in fourbetween Table). segments Combining of be needed the todata show for why,A in more our patient negative sample, average there residual was no (worseOur findings performance) on the Iowa Gamblingwas Task and deserve special explained by g (cf. Fig. 1A). When fluid intelligence is partialled attention. Some prior studies show clear decision-making deficits fluid intelligence. First, we examined the relationship betweenleft lateral,the frontal right lobe, lateral, anterior inferior and posterior medial and to Y = superior 35 in left medial and right specific association with left hemisphere lesions. One possibility patients and controls, all 10 correlationsout, were however, positive a second setassociated of tasks shows with remaining larger deficits. lesion This volumein in patients the right with anteriorventromedial region frontal deficits, manifest in consistent such deficits in the five tasks where they were found (Go–Nosubgroups.hemispheres. Mean volume of damage, and variability across is that, by comparison with many previous studies, our lesion (median = 0.30, range = 0.03 to 0.52). For patientssecond alone, set of tasks 7/10 includes(r Go–No= 0.59, go,P Proverbs,50.005, Hayling, one-tailed). Hotel Noselection significant of riskycorrelations decks (Bechara were et al., 2000; Torralva et al., go, Proverbs, Hayling, Hotel, Faux Pas). For each task, scores patients, were similar for left anterior, right anterior and left À correlations were positive (median = 0.14,sample rangeand Faux was = Pas.0.36 relatively The to data anterior, givefound some with for suggestion no left lesions anterior( that extendingr these= 0.24, addi- intoP = 0.15), the2007). right Decision-making posterior (r = deficits0.15, can exist in the absence of more for each patient and control were converted to residuals after posterior regions, but somewhat greater for right posterior. À À 0.45), while for controls alone 9/10temporaltional, were non- lobeg positive anddeficits lesions mayP bein= only associated0.25), two or of withleft our posterior the left most lateral anterior (r = patients0.19, Pgeneral= 0.20). deficits, The significant for example cor- in working memory (Clark and adjusting for fluid intelligence (Figs 3 and 4, vertical distances For each region, volumes of damage were correlated with mean Discussion (median = 0.28, range = 0.03 to 0.47). Together,incorporating(especially these right) more results frontal than lesions. 10%relation of For Brodmann for these right tests, area anterior results 44. resemble Previously, damage remained itManes, 2004). even afterAt the covarying same time, deficits in Iowa Gambling can from the regression line). Residuals in the five tasks were then performance residuals (deficits beyond theÀ prediction from fluid hasthose been of suggested Fig. 1B, implying that frontal deficits indeficits some specificin Verbal function Fluency largely may bebe seen following other kinds of prefrontal lesion (e.g. Manes suggest that, across the five tasks, deviations from the score for total lesion volume, r = 0.52, P50.01. correlated, for a total of 10 correlations between all possibleIn ourintelligence, data, the picture averaged of relations across between the five executive critical tasks) deficit as and above. overseparate and above from g. those predicted by general intelligenceÀ as mea-et al., 2002), and it has been suggested that this task depends predicted by fluid intelligence may be traced at least in part to Since scores were categorical, often with a small range, for the task pairs (see Supplementary Table). Combining the data forfluid intelligenceA more negative is both simple average and residual unanticipated. (worse performance)Certainly, the was suredFor by the the first Wechsler group of Adulttasks—Wisconsin Intelligence Card Scale Sorting (Henry Test, andon other cognitive functions besides reward coding and use, some common factor. For each patient, accordingly,Verbal we Fuency obtained and Iowa Gambling—wesubtests of found the Ineco no specific Frontal associ- Screening,including data learning, were re-examined shifting and spatial working memory (Dunn patients and controls, all 10 correlations were positiveresultsassociated show that with fluid larger intelligence lesion volume is a substantial in the right contributor anterior to region Crawford, 2004). Our results suggest that fluid intelligence, with (median = 0.30, range = 0.03 to 0.52). For patients alone, 7/10 (r = 0.59, P50.005,a mean one-tailed). residual No across significant the five correlations tasks (or were feweration for patients with particular with regionsas of appropriate prefrontal damage. using As non-parametric reviewed et altests., 2006). (Mann–Whitney In our data, Iowa U, Gambling was positively correlated frontal deficits.À For one group of ‘executive’ tasks, including the its emphasis on current problem solving, may be the more suitable correlations were positive (median = 0.14, range = 0.36 to found for left anterior(missingr = 0.24, data),P = and0.15), examined right posterior lesion (r = overlap0.15, forearlier, the sixprevious patients findings forSpearman the Wisconsin rank Card correlation, Sorting Test Kruskal–Wallis are with g. H). Once Conclusionsg was removed, were the deficit in frontal patients À Wisconsin Card Sorting Test, Verbal Fluency and Iowa Gambling,À measurecontradictory, of g in with frontal some patients studies suggesting (Duncan et specific al., 1995). deficits after became non-significant, but was still borderline. We found no 0.45), while for controls alone 9/10 were positive P = 0.25), or left posteriorwith the (r greatest= 0.19, P negative= 0.20). The value significant (i.e. greatest cor- deficit beyond the essentially identical to those based on parametric tests. differences between patients and controls can be entirely Ourdorsolateral findings lesions on the (e.g. Iowa Milner, Gambling 1963; Rezai Task andet al.,g deserve 1993), but specialevidence of selective deficit in patients whose lesions included (median = 0.28, range = 0.03 to 0.47). Together, these results relation for right anteriorprediction damage from remained fluid intelligence). even after The covarying result (Fig. 5) provides a Differences between all combined frontal patients and controls À explained by g (cf. Fig. 1A). When fluid intelligence is partialled attention.others not Some (e.g. prior Stuss studieset al., show 1983, clear 2000). decision-making Verbal Fluency deficits is the inferior medial region. Certainly these data suggest that the suggest that, across the five tasks, deviations from the score for total lesion volume, r = 0.52, P50.01. out, however, a second set of tasksÀ shows remaining deficits. This inalso patients a widely withused ventromedial test in the frontal assessment deficits, of manifest frontal functions. in consistenttask is influenced by factors in addition to a specific risky decision- predicted by fluid intelligence may be traced at least in part to Since scores were categorical, often with a small range, for the second set of tasks includes Go–No go, Proverbs, Hayling, Hotel selectionDespite ofthe risky fact that decks phonological (Bechara fluencyet al., can 2000; be impaired Torralva inet a al.,making component, which may critically depend on ventromedial some common factor. For each patient, accordingly, we obtained subtests of the Ineco Frontal Screening, data were re-examined wide range of conditions and in patients with different lesion prefrontal cortex. This risky component is perhaps more salient in and Faux Pas. The data give some suggestion that these addi- 2007). Decision-making deficits can exist in the absence of more a mean residual across the five tasks (or fewer for patients with as appropriate using non-parametric tests (Mann–Whitney U, localizations (Crawford et al., 1993; Brooks et al., 1999; Henry other patient groups, e.g. those with bilateral ventromedial tional, non-g deficits may be associated with the most anterior general deficits, for example in working memory (Clark and missing data), and examined lesion overlap for the six patients Spearman rank correlation, Kruskal–Wallis H). Conclusions were and Crawford, 2004), it has been extensively demonstrated that damage (Bechara et al., 1994) or frontal variant frontotemporal (especially right) frontal lesions. For these tests, results resemble with the greatest negative value (i.e. greatest deficit beyond the essentially identical to those based on parametric tests. Manes,frontal 2004). patients At are the more same impaired time, deficits that non-frontal in Iowa Gambling patients candementia (Torralva et al., 2007). those of Fig. 1B, implying deficits in some specific function largely prediction from fluid intelligence). The result (Fig. 5) provides a Differences between all combined frontal patients and controls be(Milner, seen following 1964; Benton, other 1968; kinds Perret, ofprefrontal 1974; Henry lesion and Crawford, (e.g. Manes For patients like ours, meanwhile, our data have strong implica- separate from g. et2004). al., 2002), Deficits and have it has been been associated suggested with a that variety this of task frontal dependstions for use and interpretation of tests such as the Wisconsin For the first group of tasks—Wisconsin Card Sorting Test, onregions, other including cognitive dorsolateral functions and besides superior reward medial coding (Stuss et and al., use,Card Sorting Test, Verbal Fluency and Iowa Gambling. To a Verbal Fuency and Iowa Gambling—we found no specific associ- including1998; Troyer learning,et al., shifting 1998). They and spatial occur with working either leftmemory or right (Dunnlarge degree, the deficits measured in such tests may not be ation with particular regions of prefrontal damage. As reviewed et al., 2006). In our data, Iowa Gambling was positively correlated earlier, previous findings for the Wisconsin Card Sorting Test are with g. Once g was removed, the deficit in frontal patients contradictory, with some studies suggesting specific deficits after became non-significant, but was still borderline. We found no dorsolateral lesions (e.g. Milner, 1963; Rezai et al., 1993), but evidence of selective deficit in patients whose lesions included others not (e.g. Stuss et al., 1983, 2000). Verbal Fluency is the inferior medial region. Certainly these data suggest that the also a widely used test in the assessment of frontal functions. task is influenced by factors in addition to a specific risky decision- Despite the fact that phonological fluency can be impaired in a making component, which may critically depend on ventromedial wide range of conditions and in patients with different lesion prefrontal cortex. This risky component is perhaps more salient in localizations (Crawford et al., 1993; Brooks et al., 1999; Henry other patient groups, e.g. those with bilateral ventromedial and Crawford, 2004), it has been extensively demonstrated that damage (Bechara et al., 1994) or frontal variant frontotemporal frontal patients are more impaired that non-frontal patients (Torralva et al., 2007). (Milner, 1964; Benton, 1968; Perret, 1974; Henry and Crawford, For patients like ours, meanwhile, our data have strong implica- 2004). Deficits have been associated with a variety of frontal tions for use and interpretation of tests such as the Wisconsin regions, including dorsolateral and superior medial (Stuss et al., Card Sorting Test, Verbal Fluency and Iowa Gambling. To a 1998; Troyer et al., 1998). They occur with either left or right large degree, the deficits measured in such tests may not be Lessons from focal lesions applied to neurodegeneration

33 Review

diagnosis without imaging or other confi rmatory test consensus criteria for bvFTD might need to be revisited results. These criteria also present diffi culties in their in the light of current research.62 application due to under-specifi cation of some features The evidence reviewed thus far indicates that no specifi c and were derived by clinical consensus prior to the cognitive profi le seems to be associated with bvFTD early publication of quantitative studies comparing cohorts in the disease, although careful cognitive assessment will with pathologically verifi ed diagnoses. Recently proposed reveal defi cits, generally in the domains of executive criteria developed by an international FTD research function and episodic memory. With disease progression, group (panel)64 build on recent work with operationalised the atrophy evolves to involve the anterior temporal defi nitions that have three levels of diagnostic certainty: regions, and the pattern of defi cits becomes less distinct possible, probable, and defi nite bvFTD. Patients qualify from other FTD subtypes, notably semantic dementia.10 for possible bvFTD on the basis of three core behavioural or cognitive features (including social disinhibition, apathy, loss of empathy, stereotypic behaviours or Panel: International consensus criteria for bvFTD alterations in eating pattern, and neuropsychological Behavioral variant Neurodegenerative disease defi cits indicative of frontal executive dysfunction). A Must be present for any FTD clinical syndrome probable diagnosis requires the same clinical features A type of frontotemporal dementia associated with behavioural change Shows progressive deterioration of behaviour and/or cognition by observation or history with evidence of progression and unequivocal neuro- imaging abnormalities. The term “defi nite” is reserved Possible bvFTD for those with neuropathology or a pathogenic gene Three of the features (A–F) must be present; symptoms should occur repeatedly, not just mutation. These new criteria (panel) are currently as a single instance: undergoing validation against neuropathological changes A Early (3 years) behavioural disinhibition by an international consortium of researchers.64 B Early (3 years) apathy or inertia C Early (3 years) loss of sympathy or empathy Neuropsychology D Early (3 years) perseverative, stereotyped, or compulsive/ritualistic behaviour Cognition E Hyperorality and dietary changes Early in the disease process, patients with bvFTD can F Neuropsychological profi le: executive function defi cits with relative sparing of perform relatively well on formal neuropsychological memory and visuospatial functions tests despite the presence of signifi cant personality and Probable bvFTD behavioural changes.65 The mini mental state examination All the following criteria must be present to meet diagnosis: is insensitive, but the Addenbrooke’s cognitive A Meets criteria for possible bvFTD examination seems to detect at least 90% of cases at B Signifi cant functional decline presentation.66 The prototypical cognitive profi le is one of C Imaging results consistent with bvFTD (frontal and/or anterior temporal atrophy on relatively preserved language and visuospatial/ CT or MRI or frontal hypoperfusion or hypometabolism on SPECT or PET) constructive abilities. Whether patients with bvFTD show executive dysfunctions remains contentious,67,68 and has Defi nite bvFTD been complicated by the inclusion of phenocopy cases. Criteria A and either B or C must be present to meet diagnosis: However, such defi cits constitute a central diagnostic A Meets criteria for possible or probable bvFTD feature of the newly proposed clinical diagnostic criteria.64 B Histopathological evidence of FTLD on biopsy at post mortem Recent evidence suggests that the combination of specifi c C Presence of a known pathogenic mutation tests (eg, digit span backward task, Hayling test of Exclusion criteria for bvFTD response inhibition, and the short version of the executive Criteria A and B must both be answered negatively; criterion C can be positive for possible and social cognition battery) might help diff erentiate bvFTD but must be negative for probable bvFTD: these cases, because results are typically abnormal in A Pattern of defi cits is better accounted for by other non-degenerative nervous system patients with true bvFTD and normal in phenocopy or medical disorders cases.68,69 B Behavioural disturbance is better accounted for by a psychiatric diagnosis The presence of severe defi cits of episodic memory has 34 C Biomarkers strongly indicative of Alzheimer’s disease or other neurodegenerative process been used as an exclusion criterion for a clinical diagnosis of bvFTD,5 although a proportion (10–15%) of patients Additional features with pathologically confi rmed FTLD present with severe A Presence of motor neuron fi ndings suggestive of motor neuron disease amnesia.21,70 A recent report has indicated that defi cits in B Motor symptoms and signs similar to corticobasal degeneration and progressive episodic memory are more common than previously supranuclear palsy reported.71 Carefully selected patients with bvFTD (ie, C Impaired word and object knowledge after excluding phenocopy cases) had similar memory D Motor speech defi cits impairments as seen in AD on tests of episodic memory, E Substantial grammatical defi cits even after accounting for disease severity.71 The criterion Criteria from Rascovsky et al.64 bvFTD=behavioural-variant frontotemporal dementia. FTD=frontotemporal of relative sparing of episodic memory compared with dementia. SPECT=single-photon emission computed tomography. FTLD=frontotemporal lobar degeneration. executive functions proposed in the recent international www.thelancet.com/neurology Vol 10 February 2011 165 Behavioral variant frontotemporal dementia Is one of the three types of frontotemporal dementia (FTD)

v v v

35 Patient video

36 Patient video

37 Behavioural changes in dysexecutive syndrome Behavioural inventory in GREFEX battery I Structured interview with caregiver

Behavioral Dysexecutive Syndrome Behavioral indices Behavioral disorders Inventory

Domains* cutoff scores at the 5% level Global hypoactivity with apathy-abulia Grouped into hypoactivity with apathy-abulia frequency x severity of 2 or 3 domains > cutoff score domain Difficulties for anticipation and initiation Disinterest and indifference Hyperactivity-distractibility-psychomotor Grouped into hyperactivity-distractibility- frequency x severity of 2 or 3 domains > cutoff score instability psychomotor instability domain

Irritability-impulsivity-aggressiveness Euphoria, emotional lability and moria Stereotyped and perseverative behavior Stereotyped and perseverative behavior frequency x severity > cutoff score Environmental dependency Environmental dependency frequency x severity > cutoff score Anosognosia-anosodiaphoria Anosognosia-anosodiaphoria frequency x severity > cutoff score Spontaneous Spontaneous confabulations frequency x severity > cutoff score Disorders of social behavior Disorders of social behavior frequency x severity > cutoff score Disorders of sexual behavior Disorders of sexual behavior frequency x severity > cutoff score Behavioral dysexecutive syndrome ≥ 3/12 domains impaired 38 Godefroy et al (2010) Ann Neurol Godefroy et al: Dysexecutive Syndrome

nitive but not behavioral dysexecutive syndrome was influenced by the underlying disease.

Discussion This study proposed criteria for dysexecutive syndrome, gave an operational definition based on easily adminis- tered tests and questionnaires, and showed that they dis- criminated between patients and controls, with a higher frequency in patients with loss of autonomy, and that both behavioral and cognitive dysexecutive syndromes were independent predictors of loss of autonomy. In addition, it provided evidence for differing patterns of cognitive disorders according to the underlying disease. This cooperative study has several limitations. It was a clinical study involving a large number of patients; the complete battery of tests and questionnaires therefore can- not be administered to all patients. A reliable informant was not available in all cases, precluding assessment of be- havioral disorders and disability. However, the subgroups did not differ from the complete sample, and the reduction of sample size due to unavailable data did not result in type II error. This study deliberately included patients suf- fering from various diseases to be representative of clinical FIGURE 1: Frequency (%) of behavioral (upper part) and populations referred for cognitive assessment. In addition, cognitive (lower part) dysexecutive disorders in patients and controls (A) and in patients (B) according to the presence of this transnosological study overcomes numerous biases loss of autonomy. related to the study of a single disease, such as effects related to associated perceptuomotor deficits (eg, stroke), Alzheimer and Parkinson diseases, (4) initiation was pre- associated cognitive deficits (eg, memory deficit in Alzhei- dominantly affected in traumatic brain injury and multi- mer disease), or the prominence of one site or type of ple sclerosis, and (5) deduction was predominantly lesions within the network subserving executive functions. affected in stroke and traumatic brain injury. The differ- The transnosological study design led us to adopt a com- Dysexecutiveences persisted (p < 0.03,syndrome all) in the subgroup with loss mon criterion for loss of autonomy based on several sever- Bothof behavioural autonomy forand initiation, cognitive generation, changes occur and planning ity scales. The loss of autonomy criterion was simple and (data not shown). This indicates that the pattern of cog- was assisted by well-validated disability scales.

Hypoactivity-Apathy Initiation

Hyperactivity- Confabulations Distractibility Inhibition Shifting

Anosognosia Stereotypies

Planning Deduction

Environmental Sexual behaviour dependency

Co-ordination Generation Social behaviour

FIGURE 2: Pattern (frequency %) of behavioral (left) and cognitive (right) disorders according to the disease. BI 5 brain injury; I 5 impairment. 39 Godefroy et al (2010) Ann Neurol

December, 2010 861 Behavioural or dysexecutive doi:10.1093/brain/awv191 BRAIN 2015: 138; 2732–2749 | 2732 2736presentations| BRAIN 2015: 138; 2732–2749 in Alzheimer’s R. Ossenkoppele et al.

Table 1 Demographic and clinical characteristics accordingThe to behavioural/dysexecutive diagnostic group variant of Behavioural/ Behavioural Alzheimer’sDysexecutive disease:Typical clinical, neuroimagingBehavioural Controls and dysexecutive variant presentation pathologicalpresentation featuresAlzheimer’s variant FTD Alzheimer’s disease Alzheimer’s disease Alzheimer’s disease disease Rik Ossenkoppele,1,2,3,4 Yolande A. L. Pijnenburg,3 David C. Perry,1 N 75 55*Brendan 29* I. Cohn-Sheehy,1 Nienke M. 58 E. Scheltens,3 Jacob 59 W. Vogel,2 Joel H. Kramer, 61 1 Annelies E. van der Vlies,3 Renaud La Joie,2 Howard J. Rosen,1 Wiesje M. van der Flier,3,5 a Age 65.8 8.5 64.7 8.8Lea T.69.2 Grinberg,8.51,6 Annemieke J. Rozemuller, 64.4 8.63 Eric J. Huang, 63.86 Bart6.8 N. M. van 63.7 Berckel,8.14 Æ Æ Bruce L. Miller,Æ 1 Frederik Barkhof,4 WilliamÆ J. Jagust,2 Philip Scheltens,Æ 3 Æ Sex (% male) 68.0 72.7William 60.7 W. Seeley1,6 and Gil D. Rabinovici 65.51,2 71.2 62.3 Education (years)b 15.5 3.1 15.7 2.3 15.7 2.7 15.8 2.5 15.4 3.2 17.3 1.9 Æ Æ Æ Æ Æ Æ MMSEc 22.7 5.6 22.5 5.4 24.6 3.3 22.5 4.1 23.7 5.4 29.4 0.7 Æ Æ Æ Æ Æ Æ CDRd 0.9 0.6 0.9 0.4A ‘frontal 0.8 variant0.3 of Alzheimer’s disease’ has been 0.9 described0.5 in patients with predominant1.1 0.7 behavioural or dysexecutive 0 0 deficits Æ Æ caused by Alzheimer’sÆ disease pathology. The descriptionÆ of this rare Alzheimer’s diseaseÆ phenotype has beenÆ limited to case GDSe 3.4 2.9 3.2 2.8reports 3.7 and small3.2 series, and many clinical, neuroimaging 2.9 and2.1 neuropathological 5.0 characteristics3.4 are not well understood.2.0 2.6 In this retrospective study, we included 55 patients with Alzheimer’s disease with a behavioural-predominant presentation (behavioural f Æ Æ Æ Æ Æ Æ NPI 14.3 16.8 15.4 17.6Alzheimer’s 12.3 disease)18.1 and a neuropathological diagnosis 7.0 of high-likelihood11.0 Alzheimer’s 21.9 disease20.0 (n=17) and/or biomarker 2.7 1.2 evidence g Æ Æ of Alzheimer’sÆ disease pathology (n=44). In addition, weÆ included 29 patients with autopsy/biomarker-definedÆ Alzheimer’sÆ disease % APOE e4 carriers 51.7 59.5with a 40.0 dysexecutive-predominant syndrome (dysexecutive 72.1 Alzheimer’s disease). We 18.9 performed structured chart reviews 16.7 to ascertain ++/+ / g clinical features. First symptoms were more often cognitive (behavioural Alzheimer’s disease: 53%; dysexecutive Alzheimer’s APOE e4 À ÀÀ 6/25/29 6/19/17disease: 2/8/15 83%) than behavioural (behavioural Alzheimer’s 14/17/12 disease: 25%; dysexecutive 0/10/43 Alzheimer’s disease: 3%). 3/7/50 Apathy was the TIV (l) 1.60 0.17 1.60 0.15most common 1.61 behavioural0.19 feature, while hyperorality 1.59 and perseverative/compulsive0.15 1.64 behaviours0.16 were less prevalent. 1.57 Fifty-two0.14 per Æ Æ cent of patientsÆ with behavioural Alzheimer’s disease metÆ diagnostic criteria for possibleÆ behavioural-variant frontotemporalÆ Autopsy-confirmed 24 17dementia. 12 Overlap between behavioural and dysexecutive 8 Alzheimer’s disease was 21 modest (9/75 patients). Sixty -per cent of patients with behavioural Alzheimer’s disease and 40% of those with the dysexecutive syndrome carried at least one APOE e4 allele. We PET/CSF biomarkers 41/22 28/18also compared 15/10 neuropsychological test performance 26/29 and brain atrophy (applying 23/23 voxel-based morphometry)- with matched aut- opsy/biomarker-defined typical (amnestic-predominant) Alzheimer’s disease (typical Alzheimer’s disease, n=58), autopsy-con- firmed/Alzheimer’s disease biomarker-negative behavioural variant frontotemporal dementia (n=59), and controls (n=61). Data are presented as mean SD unless indicated otherwise. Differences between groupsPatients were with behavioural assessed Alzheimer’s using ANOVA disease showed with worsepost memory hoc LSD scores teststhan behavioural [age, education, variant frontotemporal MMSE, dementia CDR, Æ and did not differ from typical Alzheimer’s2 disease, while executive function composite scores were lower compared to behavioural Geriatric40 Depression Scale (GDS), Neurophsychiatric Inventory (NPI) and total intracranialvariant volume frontotemporal (TIV)], dementia (sex and typical and Alzheimer’sAPOE e4 disease. status), Voxel-wise and Kruskal-Wallis contrasts between behavioural with post and hoc dysexecutiveMann- Whitney U-tests (number of APOE e4 alleles). Alzheimer’s disease patients and controls revealed marked atrophy in bilateral temporoparietalOssenkoppele regions andet only al limited (2015) atrophy Brain a in the frontal cortex. In direct comparison with behavioural and those with dysexecutive Alzheimer’s disease, patients with Dysexecutive Alzheimer’s disease 4 other groups, P 5 0.05. behavioural variant frontotemporal dementia showed more frontal atrophy and less posterior involvement, whereas patients b Controls 4 patients, P 5 0.01. with typical Alzheimer’s disease were slightly more affected posteriorly and showed less frontal atrophy (P 5 0.001 uncorrected). c Among 24 autopsied behavioural Alzheimer’s disease/dysexecutive Alzheimer’s disease patients, only two had primary co-morbid Patients 5 controls, P 5 0.001; behavioural Alzheimer’s disease + typical Alzheimer’s diseaseFTD-spectrum5 dysexecutive pathology (progressive Alzheimer’s supranuclear disease, palsy). In conclusion,P 5 0.05. behavioural Alzheimer’s disease presentations are char- d Patients 4 controls, P 5 0.001; behavioural variant FTD 4 dysexecutive Alzheimer’s disease,acterized byP 5 a milder0.01. and more restricted behavioural profile than in behavioural variant frontotemporal dementia, co-occurrence eBehavioural/dysexecutive Alzheimer’s disease + behavioural Alzheimer’s disease + dysexecutiveof memory Alzheimer’s dysfunction and disease high APOE + behaviourale4 prevalence. Dysexecutive variant FTD Alzheimer’s4 controls, disease presentedP 5 0.05; as a primarily behavioural cognitive variant FTD 4 behavioural Alzheimer’s disease/dysexecutive Alzheimer’s disease + behavioural Alzheimer’s disease + typical Alzheimer’s disease, P 5 0.01. f Behavioural/dysexecutive Alzheimer’s disease + behavioural Alzheimer’s disease + dysexecutiveReceived January Alzheimer’s 8, 2015. Revised April disease 3, 2015. Accepted + behavioural May 5, 2015. Advance variant Access publication FTD July4 3, 2015controls, P 5 0.05; behavioural ß The Author (2015). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. Alzheimer’s disease/dysexecutive Alzheimer’s disease + behavioural Alzheimer’s disease 4Fortypical Permissions, Alzheimer’s please email: [email protected] disease, P 5 0.05; behavioural variant FTD 4 dysexecutive Alzheimer’s disease + typical Alzheimer’s disease, P 5 0.05. gBehavioural/dysexecutive Alzheimer’s disease + behavioural Alzheimer’s disease + dysexecutive Alzheimer’s disease + typical Alzheimer’s disease 4 behavioural variant FTD + controls, P 5 0.05; typical Alzheimer’s disease 4 dysexecutive Alzheimer’s disease, P 5 0.05. *Nine patients met criteria for both behavioural and dysexecutive Alzheimer’s disease and were included in both groups. APOE = Apolipoprotein E; + + = e4 homozygous, + = e4 heterozygous; = e4 negative. À ÀÀ

Magnetic resonance imaging (DARTEL) was used to generate a study-specific template by aligning the grey matter images non-linearly to a common Acquisition space. Native grey and white matter images were spatially normalized to the DARTEL template using individual flow Structural MRI scans were available for 46/55 patients with fields, with modulation applied to preserve the total amount behavioural Alzheimer’s disease, 26/29 patients with dysexecu- of signal. Images were smoothed using an 8-mm full-width at tive Alzheimer’s disease, 58/58 patients with typical half-maximum isotropic Gaussian kernel. Images were in- Alzheimer’s disease, 57/59 patients with behavioural variant spected visually after each step in the processing pipeline, FTD, and 61/61 control subjects. At UCSF, T -weighted 1 and the final smoothed-modulated-warped grey matter images were acquired on a 1.5 T (Magnetom Avanto images were checked for sample homogeneity using the System/Magnetom VISION system, Siemens, n=87) or 3 T VBM8 toolbox to identify potential outliers. Next, we per- (Tim Trio, Siemens, n=51) unit. At VUMC, MRI scans formed voxel-wise contrasts between the four different patient were performed on a 1 T (Magnetom Impact, Siemens, groups (plus a combined behavioural/dysexecutive Alzheimer’s n=21), 1.5 T (Sonata, Siemens, n=23) or 3 T (SignaHDxt, disease group) and the healthy control subjects. We addition- GE Healthcare, n=66) unit. Acquisition parameters have been ally performed voxel-wise contrasts against controls for pa- published previously (Sluimer et al., 2008; Ossenkoppele et al., tients with autopsy-confirmed behavioural Alzheimer’s 2012a; Lehmann et al., 2013a; Moller et al., 2013). The pro- disease/dysexecutive Alzheimer’s disease (n=24) and for pa- portion of subjects scanned on each scanner was balanced tients with behavioural Alzheimer’s disease who were initially across groups and all imaging statistical models included scan- diagnosed with behavioural variant FTD (n=13). Finally, we ner type and acquisition site as nuisance variables. directly compared the patient groups (combined behavioural/ dysexecutive Alzheimer’s disease, typical Alzheimer’s disease Voxel-based morphometry and behavioural variant FTD). The primary models included MRI data were segmented using the New Segment toolbox age, sex, total intracranial volume, scanner type and centre as implemented in the Statistical Parametric Mapping (SPM) 8 nuisance variables. Secondary models additionally included software (Wellcome Trust Centre for Neuroimaging, Institute MMSE (as a proxy of disease severity) as a nuisance variable. of Neurology at University College London). Diffeomorphic Results are displayed at an uncorrected threshold of Anatomical Registration Through Exponentiated Lie Algebra P 5 0.001. 2738 | BRAIN 2015: 138; 2732–2749 R. Ossenkoppele et al.

The breakdown of cognitive complaints at baseline con- Results sisted of memory impairment (26/55, 47%), executive dys- function (4/55, 7%), both (21/55, 38%) or neither memory Participants nor executive dysfunction (4/55, 7%). Hypertension, de- Demographic and clinical characteristics are presented in pression, sleep disorder and traumatic brain injury were Table 1. Clinical groups consisted of mildly impaired pa- the most frequently mentioned conditions in the medical tients with mean MMSE scores ranging from 22 to 25 and history (Fig. 1B). Fifty-two per cent met international con- an average CDR of 1. Patients with behavioural sensus criteria for possible behavioural variant FTD  Alzheimer’s disease (mean age: 64.7 8.8, median: 64.1, (Rascovsky et al., 2011). The majority of patients were Æ range: 43.1–83.5) and dysexecutive Alzheimer’s disease close to the threshold of 53 of 6 core behavioural/cogni- (mean age: 69.2 8.5, median: 71.3, range: 53.7–83.5) tive symptoms required for a diagnosis of possible behav- Æ were relatively young at time of diagnosis and more often ioural variant FTD [2/6 (33%), 3/6 (22%) or 4/6 (20%), male (73% and 61%, respectively) than female. Of the Fig. 1C], and had fewer behavioural symptoms compared patients with behavioural Alzheimer’s disease and patients to patients with behavioural variant FTD. Apathy was more prominent than disinhibition and loss of empathy, with dysexecutive Alzheimer’s disease, 59.5% and 40% while hyperorality and perseverative/compulsive behaviours carried at least one APOE e4 allele, respectively. were less common (Fig. 1D). Figure 1D and the Neuropsychiatric Inventory scores also indicate that the be- Behavioural-predominant presentations of havioural profile of patients with behavioural Alzheimer’s Alzheimer’s disease disease was less profound than that of patients with behav- Patients presented initially with cognitive difficulties (53%) ioural variant FTD. Neuropsychiatric Inventory domains more often than with behavioural changes (25%, Fig. 1A). often cited in typical Alzheimer’s disease such as anxiety,

2738 | BRAIN 2015: 138; 2732–2749 R. Ossenkoppele et al.

The breakdown of cognitive complaints at baseline con- Results sisted of memory impairment (26/55, 47%), executive dys- function (4/55, 7%), both (21/55, 38%) or neither memory Participants nor executive dysfunction (4/55, 7%). Hypertension, de- Demographic and clinical characteristics are presented in pression, sleep disorder and traumatic brain injury were TableBehavioural 1. Clinical groups consistedor dysexecutive of mildly impaired pa- doi:10.1093/brain/awv191the most frequently mentioned conditionsBRAIN 2015: 138;in 2732–2749 the| 2732 medical tients with mean MMSE scores ranging from 22 to 25 and history (Fig. 1B). Fifty-two per cent met international con- anpresentations average CDR of 1. in Patients Alzheimer’s with behavioural sensus criteria for possible behavioural variant FTD  Alzheimer’s disease (mean age: 64.7 8.8, median: 64.1, (Rascovsky et al., 2011). The majority of patients were Æ range: 43.1–83.5) and dysexecutive Alzheimer’s disease Theclose behavioural/dysexecutive to the threshold of 53 of variant 6 core behavioural/cogni- of (mean age: 69.2 8.5, median: 71.3, range: 53.7–83.5) Alzheimer’stive symptoms disease: required clinical, for a diagnosis neuroimaging of possible and behav- Æ were relatively young at time of diagnosis and more often pathologicalioural variant features FTD [2/6 (33%), 3/6 (22%) or 4/6 (20%), Fig. 1C], and had fewer behavioural symptoms compared male (73% and 61%, respectively) than female. Of the Rik Ossenkoppele,1,2,3,4 Yolande A. L. Pijnenburg,3 David C. Perry,1 Brendanto I. Cohn-Sheehy, patients1 Nienke with M. behavioural E. Scheltens,3 Jacob variant W. Vogel,2 Joel FTD. H. Kramer, Apathy1 was patients with behavioural Alzheimer’s disease and patients Annelies E. van der Vlies,3 Renaud La Joie,2 Howard J. Rosen,1 Wiesje M. van der Flier,3,5 Lea T.more Grinberg,1,6 prominentAnnemieke J. Rozemuller, than disinhibition3 Eric J. Huang,6 Bart and N. M. loss van Berckel, of empathy,4 with dysexecutive Alzheimer’s disease, 59.5% and 40% Bruce L. Miller,1 Frederik Barkhof,4 William J. Jagust,2 Philip Scheltens,3 Williamwhile W. Seeley hyperorality1,6 and Gil D. Rabinovici and1,2 perseverative/compulsive behaviours carried at least one APOE e4 allele, respectively. were less common (Fig. 1D). Figure 1D and the A ‘frontalNeuropsychiatric variant of Alzheimer’s disease’ has been Inventory described in patients scores with predominant also behavioural indicate or dysexecutive that deficits the be- caused by Alzheimer’s disease pathology. The description of this rare Alzheimer’s disease phenotype has been limited to case Behavioural-predominant presentations of reports andhavioural small series, and many profile clinical, neuroimaging of patients and neuropathological with characteristics behavioural are not well understood. Alzheimer’s In this retrospective study, we included 55 patients with Alzheimer’s disease with a behavioural-predominant presentation (behavioural Alzheimer’s disease Alzheimer’s disease) and a neuropathological diagnosis of high-likelihood Alzheimer’s disease (n=17) and/or biomarker evidence of Alzheimer’sdisease disease pathology was (n= less44). In addition, profound we included 29 than patients with that autopsy/biomarker-defined of patients Alzheimer’s with disease behav- with a dysexecutive-predominant syndrome (dysexecutive Alzheimer’s disease). We performed structured chart reviews to ascertain Patients presented initially with cognitive difficulties (53%) clinical features.ioural First symptoms variant were more FTD. often cognitive Neuropsychiatric (behavioural Alzheimer’s disease: 53%; Inventory dysexecutive Alzheimer’s domains disease: 83%) than behavioural (behavioural Alzheimer’s disease: 25%; dysexecutive Alzheimer’s disease: 3%). Apathy was the more often than with behavioural changes (25%, Fig. 1A). most commonoften behavioural cited feature, whilein hyperorality typical and perseverative/compulsive Alzheimer’s behaviours disease were less prevalent.such Fifty-two as anxiety, per cent of patients with behavioural Alzheimer’s disease met diagnostic criteria for possible behavioural-variant frontotemporal dementia. Overlap between behavioural and dysexecutive Alzheimer’s disease was modest (9/75 patients). Sixty per cent of patients with behavioural Alzheimer’s disease and 40% of those with the dysexecutive syndrome carried at least one APOE e4 allele. We also compared neuropsychological test performance and brain atrophy (applying voxel-based morphometry) with matched aut- opsy/biomarker-defined typical (amnestic-predominant) Alzheimer’s disease (typical Alzheimer’s disease, n=58), autopsy-con- Figure 1 Clinical features. Frequency of (A) first symptoms reported byfirmed/Alzheimer’s patients disease and biomarker-negative caregivers, behavioural (B) variant self-reported frontotemporal dementia medical (n=59), and conditions controls (n=61). in the Patients with behavioural Alzheimer’s disease showed worse memory scores than behavioural variant frontotemporal dementia past history, (C) the number of core behavioural/cognitive symptoms met ofand did diagnostic not differ from typical criteria Alzheimer’s disease, for while behavioural executive function composite variant scores were FTD lower compared (bvFTD; to behavioural Rascovsky 41 variant frontotemporal dementia and typical Alzheimer’s disease. Voxel-wise contrasts between behavioural and dysexecutive Alzheimer’s disease patients and controls revealed marked atrophy in bilateral temporoparietal regions and only limited atrophy et al., 2011), and (D) these behavioural/cognitive features. in the frontal cortex. In direct comparison with behavioural and those with dysexecutiveOssenkoppele Alzheimer’s disease,et al (2015) patients withBrain behavioural variant frontotemporal dementia showed more frontal atrophy and less posterior involvement, whereas patients with typical Alzheimer’s disease were slightly more affected posteriorly and showed less frontal atrophy (P 5 0.001 uncorrected). Among 24 autopsied behavioural Alzheimer’s disease/dysexecutive Alzheimer’s disease patients, only two had primary co-morbid FTD-spectrum pathology (progressive supranuclear palsy). In conclusion, behavioural Alzheimer’s disease presentations are char- acterized by a milder and more restricted behavioural profile than in behavioural variant frontotemporal dementia, co-occurrence of memory dysfunction and high APOE e4 prevalence. Dysexecutive Alzheimer’s disease presented as a primarily cognitive

Received January 8, 2015. Revised April 3, 2015. Accepted May 5, 2015. Advance Access publication July 3, 2015 ß The Author (2015). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

Figure 1 Clinical features. Frequency of (A) first symptoms reported by patients and caregivers, (B) self-reported medical conditions in the past history, (C) the number of core behavioural/cognitive symptoms met of diagnostic criteria for behavioural variant FTD (bvFTD; Rascovsky et al., 2011), and (D) these behavioural/cognitive features. 2740 | BRAIN 2015: 138; 2732–2749 R. Ossenkoppele et al.

Behavioural or dysexecutive doi:10.1093/brain/awv191 BRAIN 2015: 138; 2732–2749 | 2732 presentations in Alzheimer’s 2740 | BRAIN 2015: 138; 2732–2749 R. Ossenkoppele et al. The behavioural/dysexecutive variant of Alzheimer’s disease: clinical, neuroimaging and pathological features

Rik Ossenkoppele,1,2,3,4 Yolande A. L. Pijnenburg,3 David C. Perry,1 Brendan I. Cohn-Sheehy,1 Nienke M. E. Scheltens,3 Jacob W. Vogel,2 Joel H. Kramer,1 Annelies E. van der Vlies,3 Renaud La Joie,2 Howard J. Rosen,1 Wiesje M. van der Flier,3,5 Lea T. Grinberg,1,6 Annemieke J. Rozemuller,3 Eric J. Huang,6 Bart N. M. van Berckel,4 Bruce L. Miller,1 Frederik Barkhof,4 William J. Jagust,2 Philip Scheltens,3 William W. Seeley1,6 and Gil D. Rabinovici1,2

A ‘frontal variant of Alzheimer’s disease’ has been described in patients with predominant behavioural or dysexecutive deficits caused by Alzheimer’s disease pathology. The description of this rare Alzheimer’s disease phenotype has been limited to case reports and small series, and many clinical, neuroimaging and neuropathological characteristics are not well understood. In this retrospective study, we included 55 patients with Alzheimer’s disease with a behavioural-predominant presentation (behavioural Alzheimer’s disease) and a neuropathological diagnosis of high-likelihood Alzheimer’s disease (n=17) and/or biomarker evidence of Alzheimer’s disease pathology (n=44). In addition, we included 29 patients with autopsy/biomarker-defined Alzheimer’s disease with a dysexecutive-predominant syndrome (dysexecutive Alzheimer’s disease). We performed structured chart reviews to ascertain clinical features. First symptoms were more often cognitive (behavioural Alzheimer’s disease: 53%; dysexecutive Alzheimer’s disease: 83%) than behavioural (behavioural Alzheimer’s disease: 25%; dysexecutive Alzheimer’s disease: 3%). Apathy was the most common behavioural feature, while hyperorality and perseverative/compulsive behaviours were less prevalent. Fifty-two per cent of patients with behavioural Alzheimer’s disease met diagnostic criteria for possible behavioural-variant frontotemporal dementia. Overlap between behavioural and dysexecutive Alzheimer’s disease was modest (9/75 patients). Sixty per cent of patients with behavioural Alzheimer’s disease and 40% of those with the dysexecutive syndrome carried at least one APOE e4 allele. We Figure 2 Neuropsychological performance. Composite z-scores for (A) memory,also (B compared) executive neuropsychological functions, test performance (C) and language brain atrophy and (applying (D voxel-based) visuo- morphometry) with matched aut- opsy/biomarker-defined typical (amnestic-predominant) Alzheimer’s disease (typical Alzheimer’s disease, n=58), autopsy-con- spatial functions. Differences between groups were assessed using ANOVA. *All patient groupsfirmed/Alzheimer’s5 controls, disease biomarker-negativeP 5 0.001; behavioural **behavioural variant frontotemporal Alzheimer’s dementia (n=59), and controls (n=61). Patients with behavioural Alzheimer’s disease showed worse memory scores than behavioural variant frontotemporal dementia disease (bAD; P 5 0.001), typical Alzheimer’s disease (tAD; P 5 0.001), behavioural Alzheimer’sand did not differ disease/dysexecutive from typical Alzheimer’s disease, while Alzheimer’s executive function disease composite scores (bAD/ were lower compared to behavioural 42 variant frontotemporal dementia and typical Alzheimer’s disease. Voxel-wise contrasts between behavioural and dysexecutive deAD; P 5 0.05) and behavioural variant FTD (bvFTD; P 5 0.05) 5 dysexecutive Alzheimer’sAlzheimer’s disease; disease patients ***behavioural and controls revealed marked Alzheimer’s atrophy in bilateral disease temporoparietal regions and only limited atrophy in the frontal cortex. In direct comparison with behavioural and those with dysexecutiveOssenkoppele Alzheimer’s disease,et al (2015) patients withBrain (P 5 0.01) and typical Alzheimer’s disease (P 5 0.001) 5 behavioural variant FTD; #dysexecutivebehavioural variant Alzheimer’s frontotemporal dementia disease showed5 moretypical frontal atrophy Alzheimer’s and less posterior dis- involvement, whereas patients ease (P 5 0.01) and behavioural variant FTD (P 5 0.01) and behavioural Alzheimer’s disease/dysexecutivewith typical Alzheimer’s disease were Alzheimer’s slightly more affected disease posteriorly and and showed behavioural less frontal atrophy (P 5 0.001 uncorrected). Among 24 autopsied behavioural Alzheimer’s disease/dysexecutive Alzheimer’s disease patients, only two had primary co-morbid Alzheimer’s disease 5 typical Alzheimer’s disease (P 5 0.05) and behavioural variant FTDFTD-spectrum (P 5 0.05). pathology (progressive supranuclear palsy). In conclusion, behavioural Alzheimer’s disease presentations are char- acterized by a milder and more restricted behavioural profile than in behavioural variant frontotemporal dementia, co-occurrence of memory dysfunction and high APOE e4 prevalence. Dysexecutive Alzheimer’s disease presented as a primarily cognitive behavioural variant FTD revealed a clearly distinct atrophy anterior cingulate, orbitofrontal cortex, middle and super- Received January 8, 2015. Revised April 3, 2015. Accepted May 5, 2015. Advance Access publication July 3, 2015 ß The Author (2015). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. pattern with posterior involvement in behavioural/dysexecu- ior frontalFor Permissions, gyrus) please email: compared [email protected] to patients with typical tive Alzheimer’s disease and anterior involvement in behav- Alzheimer’s disease, but markedly less than patients with ioural variant FTD, which largely survived P 5 0.05 FWE behavioural variant FTD (Fig. 4). Also, the frequencies correction. Additional adjustment for disease severity (using were highest in regions typically affected in Alzheimer’s MMSE) essentially did not change the results (Supplementary disease such as posterior cingulate, precuneus and lateral Fig. 4). Supplementary Table 2 includes the coordinates of temporoparietal regions. Patients with dysexecutive local maxima and their anatomical labels, T-values, P-values Figure 2 Neuropsychological performance. Composite z-scoresAlzheimer’s for (A) memory, disease (B) executive showed functions, more frequent (C) language relative and (D involve-) visuo- and clusterspatial sizes functions. of all Differences voxel-wise between contrasts groups described were assessed above. using ANOVA.ment *All of patient the frontal groups 5 andcontrols, parietalP 5 0.001; cortex **behavioural than patients Alzheimer’s with disease (bAD; P 5 0.001), typical Alzheimer’s disease (tAD; P 5 0.001),behavioural behavioural Alzheimer’s Alzheimer’s disease/dysexecutive disease, whereas Alzheimer’s patients disease with (bAD/ be- FrequencydeAD; mapsP 5 0.05) of and brain behavioural atrophy variant FTD (bvFTD; P 5 0.05) 5 dysexecutivehavioural Alzheimer’s Alzheimer’s disease; disease ***behavioural showed Alzheimer’s more (medial) disease tem- To further(P 5 assess0.01) and the typical relative Alzheimer’s involvement disease (P 5 of0.001) the5 frontalbehavioural variantporal FTD; lobe #dysexecutive involvement Alzheimer’s than patients disease 5 typical with Alzheimer’s dysexecutive dis- ease (P 0.01) and behavioural variant FTD (P 0.01) and behavioural Alzheimer’s disease/dysexecutive Alzheimer’s disease and behavioural cortex we computed5 frequency maps showing5 for each Alzheimer’s disease. Alzheimer’s disease 5 typical Alzheimer’s disease (P 5 0.05) and behavioural variant FTD (P 5 0.05). voxel the proportion of patients with suprathreshold W- scores (W 5 1). Visual inspection of the W-score maps À indicatebehavioural that patients variant in both FTD behavioural revealed a clearly Alzheimer’s distinct dis- atrophy Neuropathologyanterior cingulate, orbitofrontal cortex, middle and super- ease andpattern dysexecutive with posterior Alzheimer’s involvem diseaseent in behavioural/dysexecu- groups showed Neuropathologicalior frontal gyrus) data compared were available to patients for 12 patients with typical with more frequenttive Alzheimer’s involvement disease in and frontal anterior cortical involvement regions in (i.e. behav- behaviouralAlzheimer’s Alzheimer’s disease, but disease, markedly seven less with than dysexecutive patients with ioural variant FTD, which largely survived P 5 0.05 FWE behavioural variant FTD (Fig. 4). Also, the frequencies correction. Additional adjustment for disease severity (using were highest in regions typically affected in Alzheimer’s MMSE) essentially did not change the results (Supplementary disease such as posterior cingulate, precuneus and lateral Fig. 4). Supplementary Table 2 includes the coordinates of temporoparietal regions. Patients with dysexecutive local maxima and their anatomical labels, T-values, P-values Alzheimer’s disease showed more frequent relative involve- and cluster sizes of all voxel-wise contrasts described above. ment of the frontal and parietal cortex than patients with behavioural Alzheimer’s disease, whereas patients with be- Frequency maps of brain atrophy havioural Alzheimer’s disease showed more (medial) tem- To further assess the relative involvement of the frontal poral lobe involvement than patients with dysexecutive cortex we computed frequency maps showing for each Alzheimer’s disease. voxel the proportion of patients with suprathreshold W- scores (W 5 1). Visual inspection of the W-score maps À indicate that patients in both behavioural Alzheimer’s dis- Neuropathology ease and dysexecutive Alzheimer’s disease groups showed Neuropathological data were available for 12 patients with more frequent involvement in frontal cortical regions (i.e. behavioural Alzheimer’s disease, seven with dysexecutive Behavioural/dysexecutive Alzheimer’s disease BRAIN 2015: 138; 2732–2749 | 2741

Behavioural/dysexecutive Alzheimer’s disease BRAIN 2015: 138; 2732–2749 | 2741

Behavioural or dysexecutive doi:10.1093/brain/awv191 BRAIN 2015: 138; 2732–2749 | 2732 presentations in Alzheimer’s Behavioural/dysexecutive Alzheimer’s disease BRAIN 2015: 138; 2732–2749 | 2741

Behavioural AD TheBehavioural behavioural/dysexecutivevariant FTD variant of Alzheimer’s disease: clinical, neuroimaging and pathological features

Rik Ossenkoppele,1,2,3,4 Yolande A. L. Pijnenburg,3 David C. Perry,1 Brendan I. Cohn-Sheehy,1 Nienke M. E. Scheltens,3 Jacob W. Vogel,2 Joel H. Kramer,1 Annelies E. van der Vlies,3 Renaud La Joie,2 Howard J. Rosen,1 Wiesje M. van der Flier,3,5 Dysexecutive AD LeaTypical T. Grinberg, AD1,6 Annemieke J. Rozemuller,3 Eric J. Huang,6 Bart N. M. van Berckel,4 Figure 3 Voxel-wise comparisonsBruce of grey L. Miller, matter1 Frederik volumes Barkhof, between4 William healthy J. controls Jagust,2 Philip and the Scheltens, different3 diagnostic groups. 1,6 1,2 Contrasts were adjusted for age, sex,William total intracranial W. Seeley volume,and scanner Gil D. Rabinovici type and centre. Results are superimposed on the SPM8 single-subject template (left) and the study-specific DARTEL template (right), and displayed at P 5 0.001 uncorrected. Supplementary Table 1 includes the coordinates of local maxima and their anatomical labels, T-values, P-values and cluster sizes. A ‘frontal variant of Alzheimer’s disease’ has been described in patients with predominant behavioural or dysexecutive deficits caused by Alzheimer’s disease pathology. The description of this rare Alzheimer’s disease phenotype has been limited to case Behavioural + Dysexecutive AD reports and small series, and many clinical, neuroimaging and neuropathological characteristics are not well understood. In this retrospective study, we included 55 patients with Alzheimer’s disease with a behavioural-predominant presentation (behavioural Alzheimer’s disease and five withAlzheimer’s mixed disease) behavioural/dysex- and a neuropathological diagnosiswith of behavioural high-likelihood Alzheimer’s variant disease FTD (n=17) and and/or the biomarker other evidence with mixed of Alzheimer’s disease pathology (n=44). In addition, we included 29 patients with autopsy/biomarker-defined Alzheimer’s disease ecutive Alzheimer’s disease. Thewith mean a dysexecutive-predominant interval from time syndrome of (dysexecutivebehavioural Alzheimer’s disease).variant We performed FTD and structured progressive chart reviews to ascertain supranuclear diagnosis to autopsy was 60.9 clinical23.8 features. months. First symptoms All patients were more oftenpalsy. cognitive Several (behavioural co-pathologies Alzheimer’s disease: were 53%; observeddysexecutive Alzheimer’s in our neuro- Ædisease: 83%) than behavioural (behavioural Alzheimer’s disease: 25%; dysexecutive Alzheimer’s disease: 3%). Apathy was the with behavioural/dysexecutive Alzheimer’s disease met pathological sample: cerebral amyloid angiopathy in 14/22 most common behavioural feature, while hyperorality and perseverative/compulsive behaviours were less prevalent. Fifty-two per NIA-Reagan neuropathologicalcent criteria of patients with (Hyman behavioural andAlzheimer’s disease(64%, met six diagnostic mild, criteria seven for possible moderate behavioural-variant and one frontotemporal severe), cerebro- Trojanowski, 1997) for high-likelihooddementia. Overlap Alzheimer’s between behavioural disease, and dysexecutivevascular Alzheimer’s disease disease in was 6/24 modest (25%, (9/75 patients). see Sixty legend per cent of of patients Table 2 for Figure 3 Voxel-wise comparisonswith behaviouralof grey matter Alzheimer’s volumes disease and between 40% of those healthy with the dysexecutive controls syndrome and the carried different at least one diagnosticAPOE e4 allele. groups We . with frequent neuritic plaque scores (Mirra et al., 1991) specification), Lewy body disease in 10/24 (42%, four Contrasts were adjusted for age, sex,also total compared intracranial neuropsychological volume, test scanner performance type and and brain centre. atrophy Results (applying are voxel-based superimposed morphometry) on the with SPM8 matched single-subject aut- and Braak stage V (n=4) or VIopsy/biomarker-defined (n=20) for neurofibrillary typical (amnestic-predominant)amygdala-predominant, Alzheimer’s disease (typical Alzheimer’s four transitional disease, n=58), limbic autopsy-con- and two template (left) and the study-specificfirmed/Alzheimer’s DARTEL template disease (right biomarker-negative), and displayed behavioural at P 5 variant0.001 frontotemporal uncorrected. dementia Supplementary (n=59), and Table controls 1 includes (n=61). the tanglescoordinates (Braak of local and maxima Braak, and 1991; theirPatients Braak anatomical withet behavioural al. labels,, 2006). Alzheimer’s T-values, For diseaseP-values showeddiffuse and worse cluster neocortical),memory sizes. scores than behaviouralargyrophilic variant frontotemporal grain disease dementia in 4/9 eight patients, Thal amyloid-b andplaque did not differstage from (Thal typical Alzheimer’set al., disease,(44%, while executive all limbic), function composite and TARDBP scores were lower in compared 1/12 patients to behavioural (8%, un- 43 2002) was assessed. All had stagevariant V frontotemporal and thus dementia classified and typical Alzheimer’sclassifiable-limbic). disease. Voxel-wise Additionally, contrasts between behavioural argyrophilic and dysexecutive thorny astro- Alzheimer’s disease patients and controls revealed marked atrophy in bilateral temporoparietal regions and only limited atrophy for ‘high’ Alzheimer’s disease neuropathologicalin the frontal cortex. In change direct comparison ac- withcyte behavioural clusters and those (Munoz with dysexecutiveetOssenkoppele al., 2007) Alzheimer’s were disease,et al observed (2015) patients withBrain in three cordingAlzheimer’s to recently disease proposed and five with NIA-AAbehavioural mixed guidelines variant behavioural/dysex- frontotemporal (Montine dementia showedpatients,with more behavioural frontal and atrophyone patientvariant and less posterior had FTD a involvement, large and the pituitary whereas other patients adenoma. with mixed with typical Alzheimer’s disease were slightly more affected posteriorly and showed less frontal atrophy (P 5 0.001 uncorrected). etecutive al., 2012). Alzheimer’s Of all patients, disease. The twoAmong mean had 24 mixed interval autopsied dementia behavioural from time Alzheimer’s with of disease/dysexecutivebehavioural Alzheimer’s variant disease FTD patients, and only two progressive had primary co-morbid supranuclear Alzheimer’sdiagnosis to disease autopsy and was progressive 60.9 FTD-spectrum23.8 supranuclear months. pathology All (progressive palsy patients assupranuclearpalsy. palsy). In Several conclusion, co-pathologies behavioural Alzheimer’s were disease observed presentations arein char- our neuro- co-primary neuropathological diagnosis.Æacterized by a Both milder and patients more restricted met behavioural profile than in behavioural variant frontotemporal dementia, co-occurrence with behavioural/dysexecutiveof Alzheimer’s memory dysfunction disease and high APOE met e4 prevalence.Discussionpathological Dysexecutive sample: Alzheimer’s cerebral disease presented amyloid as a primarilyangiopathy cognitive in 14/22 formalNIA-Reagan criteria for neuropathological possible behavioural criteria variant (Hyman FTD during and (64%, six mild, seven moderate and one severe), cerebro- In this retrospective study we assessed the clinical, neuro- Figure 3 Voxel-wise comparisons of grey matter volumes between healthy controls andlifeTrojanowski, the and different additionally 1997) diagnostic for met high-likelihood groups our. criteria Alzheimer’s for dysexecutive disease, vascular disease in 6/24 (25%, see legend of Table 2 for Alzheimer’s disease. One patientReceived was January clinically 8, 2015. Revised diagnosed April 3, 2015. Accepted Maypsychological, 5, 2015. Advance Access publication morphological July 3, 2015 and neuropathological Contrasts were adjusted for age, sex, total intracranial volume, scanner type and centre. Results arewith superimposed frequent on neuritic the SPM8 plaque single-subject scoresß The Author (Mirra (2015). Publishedet by al. Oxford, 1991) University Press on behalfspecification), of the Guarantors of Brain. Lewy All rights bodyreserved. disease in 10/24 (42%, four template (left) and the study-specific DARTEL template (right), and displayed at P 5 0.001 uncorrected.and Supplementary Braak stage Table V (n= 1 includes4) or VI theFor ( Permissions,n=20) please for email: neurofibrillary [email protected] amygdala-predominant, four transitional limbic and two coordinates of local maxima and their anatomical labels, T-values, P-values and cluster sizes. tangles (Braak and Braak, 1991; Braak et al., 2006). For diffuse neocortical), argyrophilic grain disease in 4/9 eight patients, Thal amyloid-b plaque stage (Thal et al., (44%, all limbic), and TARDBP in 1/12 patients (8%, un- 2002) was assessed. All had stage V and thus classified classifiable-limbic). Additionally, argyrophilic thorny astro- for ‘high’ Alzheimer’s disease neuropathological change ac- cyte clusters (Munoz et al., 2007) were observed in three Alzheimer’s disease and five with mixed behavioural/dysex- with behavioural variantcording FTD to and recently the other proposed with NIA-AA mixed guidelines (Montine patients, and one patient had a large pituitary adenoma. ecutive Alzheimer’s disease. The mean interval from time of behavioural variant FTDet al., and2012). progressive Of all patients, supranuclear two had mixed dementia with diagnosis to autopsy was 60.9 23.8 months. All patients palsy. Several co-pathologiesAlzheimer’s were disease observed and in progressive our neuro- supranuclear palsy as Æ with behavioural/dysexecutive Alzheimer’s disease met pathological sample: cerebralco-primary amyloid neuropathological angiopathy in diagnosis. 14/22 Both patients met Discussion NIA-Reagan neuropathological criteria (Hyman and (64%, six mild, sevenformal moderate criteria and for one possible severe), behavioural cerebro- variant FTD during Trojanowski, 1997) for high-likelihood Alzheimer’s disease, vascular disease in 6/24life (25%, and additionally see legend of met Table our 2 criteria for for dysexecutive In this retrospective study we assessed the clinical, neuro- with frequent neuritic plaque scores (Mirra et al., 1991) specification), Lewy bodyAlzheimer’s disease disease. in 10/24 One (42%, patient four was clinically diagnosed psychological, morphological and neuropathological and Braak stage V (n=4) or VI (n=20) for neurofibrillary amygdala-predominant, four transitional limbic and two tangles (Braak and Braak, 1991; Braak et al., 2006). For diffuse neocortical), argyrophilic grain disease in 4/9 eight patients, Thal amyloid-b plaque stage (Thal et al., (44%, all limbic), and TARDBP in 1/12 patients (8%, un- 2002) was assessed. All had stage V and thus classified classifiable-limbic). Additionally, argyrophilic thorny astro- for ‘high’ Alzheimer’s disease neuropathological change ac- cyte clusters (Munoz et al., 2007) were observed in three cording to recently proposed NIA-AA guidelines (Montine patients, and one patient had a large pituitary adenoma. et al., 2012). Of all patients, two had mixed dementia with Alzheimer’s disease and progressive supranuclear palsy as co-primary neuropathological diagnosis. Both patients met Discussion formal criteria for possible behavioural variant FTD during life and additionally met our criteria for dysexecutive In this retrospective study we assessed the clinical, neuro- Alzheimer’s disease. One patient was clinically diagnosed psychological, morphological and neuropathological Downloaded from http://jnnp.bmj.com/ on March 20, 2017 - Published by group.bmj.com

Hypotheses had at least 2 years of prominent impairment of language or We tested three hypotheses about EF and DB in groups of speech without other relevant symptoms, and although the diag- patients likely to have mild (ie, CBS and PPA) or substantial dif- noses were conferred prior to 2011, they were consistent with ficulty (ie, bvFTD). The first hypothesis involves criterion valid- the more recent criterion.29 Case definition for CBS was pro- ity, namely, that these measures will be sensitive to greater gressive, predominantly lateralised ideomotor apraxia and/or pathology in bvFTD. The second hypothesis is that EF and DB non-DOPA-responsive extrapyramidal motor dysfunction (limb are distinct but related constructs, and that this property should dystonia or rigidity) with cortical sensory loss (astereognosis, be invariant across disorders. We hypothesised that there would agraphesthesia) and characteristic patterns of atrophy on MRI.30 be overlapping and distinct neuroanatomical correlates of EF Dementia severity was assessed by Mattis Dementia Rating and DB in the prefrontal cortex. The third hypothesis examines Scale-2.19 Patients were excluded if they lacked a participating incremental utility in the conjoint use of EF and DB measures in caregiver, they were diagnosed with another neurodegenerative distinguishing bvFTD from CBS and PPA (ie, the addition of a disorder or their behaviour precluded neuropsychological second data modality yields diagnostic and descriptive testing. advantages). Demographics MATERIALS AND METHODS Data from 243 patients were included: 124 with bvFTD, 34 Participants and clinical testing with PPA and 85 with CBS. Participants were on average Participants were derived from a cohort of 243 patients, 62 years of age (SD=8.66) and had 15 years of education enrolled in studies at the National Institute of Neurological (SD=2.70). Ninety per cent were right-handed, and 96.3% Disorders and Stroke (NINDS) fromDownloaded 2001 to from 2009. http://jnnp.bmj.com/ Patients onwere March Caucasian 20, 2017 - Published (table 1 byfor group.bmj.com all characteristics). One-way analyses were recruited nationally via advertisement and required to have of variance (ANOVAs) revealed that bvFTD participants were a clinical diagnosis of either FTD or CBS from a neurologist younger than CBS participants (F2, 239=11.23, p<0.001), and CBS participants were less educated than PPA participants (F experienced withFigure both 1 Con disorders.firmatory factor Diagnoses analysis were either con- 2, firmed, or themodels appropriate in SEM. diagnosis was conferred, at NINDS 239=3.84, p<0.05). Groups did not differ in years since by an experienced behavioural neurologist (EW) and neuro- symptom onset or gender according to an ANOVA and χ2 ana- psychologist ( JHG) via consensus following evaluation with lysis, respectively. standardised neuroimaging, genetic testing and core clinical neuropsychological and neurological examinations. Data availability and imaging subsample Case definitions of bvFTD and PPA were based on Protocol completeness varied based on procedure tolerance and Lund-Manchester Group consensus criteria.Downloaded4 Although from more http://jnnp.bmj.com/scan quality. Two hundred on andMarch four participants20, 2017 had - Published image data, by group.bmj.com recent criteria have since been established,5 the previous criteria and 182 could be automatically reconstructed. Twelve studies were more stringent, with relatively lower sensitivity (52%) and were discarded due to insufficient quality and poor grey/white higher specificity (estimated 90 100%). In bvFTD, lead symp- matter segmentation leaving 170 studies. FrSBe protocols were toms were progressive dysexecutive and behavioural abnormal- available on 151 participants, and DKEFS composites on 120 itiesBehavioural with early loss of insight / noteddysexecutive by caregivers. PPA patientssyndromes(table 2). For anyin analysis,other the maximum number of available TransdiagnosticTable 2 Cognitiveapproach and across behavioural three disorders characteristics Table 1 Descriptive statistics of participant characteristicsOverall bvFTD PPA CBS Overall M (SD)bvFTD | n (%) M (SD)PPA | n (%) MCBS (SD) | n (%) M (SD) | n (%) F p Value

Sample size 243 (100.0%) 124 (51.0%) 34 (14.0%) 85 (35.0%) GenderImage availability 170 (70%) 77 (62%) 25 (74%) 68 (80%) DKEFSMale Delis-Kaplan Executive126 (51.9%) Function System 66 (53.2%) 15 (44.1%) 45 (52.9%) Female 117 (48.1%) 58 (46.8%) 19 (55.9%) 40 (47.1%) RaceVerbal fluency 4.52 (3.3) | 168 (69) 4.07 (3.5) | 74 (59) 4.83 (2.7) | 18 (53) 4.88 (3.2) | 76 (89) 1.24 ns WhiteTower Test234 (96.3%) 5.27121 (4.1) (97.6%) | 155 (63) 4.1632 (94.1%) (3.6) | 68 (55)81 7.77 (95.3%) (5.0) | 22(65) 5.58 (3.9) | 65(76) 7.34 0.001* Black 4 (1.6%) 1 (0.8%) 1 (2.9%) 2 (2.4%) AsianSorting Test1 (0.4%) 5.90 (0.0%) (3.6) | 134 (55) 4.750 (0.0%) (3.3) | 55 (44)1 5.69 (1.2%) (3.4) | 16 (66) 6.95 (3.6) | 63 (74) 6.11 0.003† PacificEF 3Islander composite1 (0.4%) 5.640 (0.0%) (3.0) | 120 (49) 4.780 (0.0%) (3.0) | 47 (38)1 (1.2%) 6.5 (2.4) | 14 (41) 5.9 (2.9) | 59 (69) 3.36 0.04 Other 3 (1.2%) 2 (1.6%) 1 (2.9%) 0 (0.0%) EthnicityFrSBe Frontal Systems Behavioural Scale NonApathy hispanic 239 (98.4%) 89.59121 (26.30) (97.6%) | 151 (62) 103.7733 (97.1%) (21.16) | 79 (63) 76.1085 (100.0%) (26.97) | 20 (59) 73.23 (20.57) | 52 (61) 35.19 0.001* Hispanic 3 (1.2%) 2 (1.6%) 1 (2.9%) 0 (0.0%) UnknownDisinhibition1 (0.4%) 72.541 (0.8%) (25.23) | 151 (62) 83.670 (0.0%) (24.61) | 79 (63) 63.600 (0.0%) (19.68) | 20 (59) 59.08 (19.84) | 52 (61) 20.63 0.001* HandednessExecutive dysfunction 85.85 (23.23) | 151 (62) 97.99 (16.94) | 79 (63) 75.15 (24.15) | 20 (59) 71.54 (21.17) | 52 (61) 32.29 0.001* Right handed 215 (88.5%) 111 (89.5%) 27 (79.4%) 77 (90.6%) LeftTotal handed 26 (10.7%) 89.2512 (9.7%) (25.92) | 151 (62) 103.847 (20.6%) (19.6) | 79 (63) 76.757 (8.2%) (26.25) | 20 (59) 71.88 (20.87) | 52 (61) 40.40 0.000‡ 44 DRSAmbidextrous 2 2 (0.8%) 1 (0.8%) 0 (0.0%) 1 (1.2%) Age M=62.07 (±8.66) M=59.82* (±8.60) M=62.06 (±8.51) GanslerM=65.38et al (2017) (±7.78) JNNP Table 3 Model fit indices of confirmatory factor analyses Table 4 β weights and estimated correlation coefficients by group EducationAttentionM=15.03 (±2.70) 30.90M=15.25 (7.06) (±2.91) | 225 (93) 30.62M=16.41 (7.48) (±1.83) | 113 (91)M=14.92 30.0† (7.81)(±2.57) | 33 (97) 31.68 (6.06) | 79 (93) 0.84 ns Model number df C /df CFI RMSEA AIC from model 4 Disease duration M=4.56 (±2.98)min M=4.58 (±3.32) M=4.01 (±1.66) M=4.75 (±2.86) Initiation/perseveration 22.14 (9.92) | 223 (92) 20.99 (10.19)bvFTD | 113 (91) CBS 20.06PPA (9.33) | 33 (97) 24.73 (9.35) | 77 (91) 4.22 0.02† *The bvFTD group1 significantly different 9 from the 11.76 CBS group at 0.75 p<0.001. 0.21 141.83 †TheConstruction CBS group significantly2 different from 9 the PPA 6.78 group at p<0.05. 0.87 0.153.97 (2.29) 97.04 | 222DKEFS (91) 4.64 (2.95) | 113 (91) 5.45 (1.12) | 33 (97) 2.34 (2.21) | 76 (89) 42.73 0.000§ 3 8 4.88 0.92 0.13 77.09 Letter fluency 0.76 0.47 0.39, ns Conceptualisation4 24 1.94 0.93 0.0628.99 (9.82) 160.59 | 225 (93)Tower Test 26.08 (10.90) |0.50 113 (91) 0.87 26.12 0.55, (9.65) ns | 33 (97) 34.35 (4.85) | 79 (93) 21.45 0.000§ Memory 5 11 2.39 0.93 0.1017.67 (6.34) 55.66 | 225 (93)Sorting Test 16.43 (6.89) | 1130.69 (91) 0.79 15.67 0.50, (5.61) ns | 33 (97) 20.27 (4.86) | 79 (93) 11.39 0.000§ 6 11 1.71 0.93 0.09 66.84 FrSBe Total α values associated with specific β weights are <0.001 unless103.65 otherwise indicated (30.06) in | 223Apathy (92) 98.70 (33.04) |0.55 113 (91) 0.76 97.58 0.85 (28.11) | 33 (97) 113.53 (23.38) | 77 (91) 6.69 0.002§ table 3. Disinhibition 0.64 0.51 0.74 Boston Naming Test 39.12 (16.90) | 204Executive (84) dysfunction 40.77 (15.70) |1.0 110 (89) 1.0 21.411.0 (19.30) | 32 (94) 45.32 (10.59) | 82 (96) 28.27 0.000¶ Token Test 80.11 (19.90) | 182Estimated (75) correlation 78.57 coefficient (20.40) of EF |0.04, 102 ns (82) 0.55, p<005 74.13 0.72, (20.03) ns | 30 (88) 86.84 (17.16) | 50 (58) 4.70 0.01§ 76%, 81% and 85%. For bvFTD, classification rates across steps and DB latent variables Visual scanningwere 90%, from 89% the and Trail 91%. Making Adding test modalities 4.95 advanced (3.91) CBS | 165 (68)α values associated 6.06 with specific (4.11)β weights | 77 are <0.001 (62) unless otherwise 5.74 indicated (3.66) in | 23(68) 3.35 (3.22) l 65 (76) 10.36 0.000§ classification, but there was no advancement for bvFTD beyond table 4. *SignifiesDB the data. bvFTD group is significantly different from the PPA group. †Signifies theNinety bvFTD bvFTD group (n=72) is significantly and PPA (n=18) different participants from had the DB, CBS group.13% improvement. For bvFTD participants, per cent correct ‡ EF and visual confrontation naming data (Boston Naming Test). classification varied from 94% to 96% to 97%, and for PPA par- SignifiesUsing the bvFTD the base group rates, is a diagnostician significantly would different have from80% accuracy. the CBS andticipants PPA groups. advanced from 61% to 67% to 78%. Classification §SignifiesBinary the CBS logistic group regression is significantly yielded overall different classi fromfication the rates bvFTD at andimproved PPA groups. stepwise for PPA, while the EF step contributed little ¶Signifiessteps PPA 1 is (BNT), significantly 2 (DB) differentand 3 (EF) from of 88%, the 90%bvFTD and and 93%, CBS a groups.to identifying bvFTD.

protocols was used. Neuroimaging data were analysed from a available online (http://surfer.nmr.mgh.harvard.edu/). Following subsample of participants (n=110) who had valid DKEFS and automated cortical reconstruction, all brains were visually FrSBe data. This ‘imaging’ subsample included participants with inspected for quality control. Abnormalities in intensity normal- bvFTD (n=53), CBS (n=41) and PPA (n=16) without any isation, resulting in poor white matter segmentation, were fixed missing data. Independent samples t-tests and χ2 analyses con- by manual placement of control points. Cortical thickness was firmed that demographics from this subsample did not signifi- calculated as the closest distance from the grey/white matter cantly differ from the larger cohort by age, education, race, boundary to the grey boundary at each surface vertex.31 handedness or disease duration. Parcellation of the cerebral cortex into gyral- and sulcal-based regions of interest was also performed, from which average cor- Standard protocol approvals, registrations and patient tical thickness measurements were calculated.32 consents Neuroanatomical correlates were identified using Qdec, in Patients gave assent for the study. Study and consent procedure FreeSurfer. Statistical surface maps were generated using general were approved by the NINDS IRB. linear modelling to explore the effects of the EF3 composite and the Total Caregiver score from the FrSBe on cortical thick- MRI protocol ness at each surface vertex. Correction for multiple comparisons Brain scans were obtained on 3 scanners of 2 field strengths was performed by Monte Carlo simulation at a statistical thresh- over the course of the study, 137 on a General Electric 1.5 T old of p<0.01. Field strength of the scanner and years of educa- with standard quadrature head coil, 2 on a Philips 1.5 T and 43 tion were added as covariates. on a Philips 3 T magnetic resonance imager. The GE imager (GE Medical Systems, Milwaukee, Wisconsin, USA) T1-weighted Assessment of EF, DB and other relevant characteristics spoiled gradient-echo sequence was used to generate 124 con- The Delis-Kaplan Executive Function System (DKEFS)33 mea- tiguous 1.5 mm-thick axial sections (repetition time, 6.1 ms; sured EF. A composite consisted of the average of three DKEFS echo time, minimum full; flip angle, 20°; field of view, 240 mm; age-adjusted subscale scores (ie, EF3), which were more robustly 124 sections; section thickness, 1.5 mm; matrix size, available and validated for coherence by factor analysis, namely, 256×256×124). 3T scan parameters were TE/TR=3/6.5 ms, letter fluency from the Verbal Fluency test, the achievement flip angle 8°, inplane resolution=256×256 voxels, 1 mm slice weighted scaled score of the Tower Test and the combined thickness, 140 180 slices, field of view=240 mm. scaled score of the Sorting Test. The FrSBe,22 which yields age and education-adjusted t-scores, was used to measure DB. FrSBe Quantitative digital image analysis caregiver responses based on patient status within the prior Automated cortical reconstruction, volumetric segmentation, month were used. For each statistical analysis, the maximum quantitative analyses and visualisation were performed using the number of available protocols was used, resulting in varying FreeSurfer software. Full documentation and download is number of participants per analysis depending on the measure Downloaded from http://jnnp.bmj.com/ on March 20, 2017 - Published by group.bmj.com

Downloaded from http://jnnp.bmj.com/ on March 20, 2017 - Published by group.bmj.com Figure 2 Neural correlates of executive function (EF) and Downloaded from http://jnnp.bmj.com/ on March 20, 2017 - Published by group.bmj.com dysexecutive behaviour (DB). Red regions indicate unique neural Figurecorrelates 2 ofNeural EF, including correlates left of executivepremotor andfunction supplemental (EF) and motorDownloaded from http://jnnp.bmj.com/Behavioural on March 20, 2017 - Published / dysexecutive by group.bmj.com syndromes in other dementias Figuredysexecutiveregions, 2 leftNeural lateral behaviour correlates orbital (DB). cortex of Red and executiveregionsbilateral indicate middle function temporalunique (EF) and neural gyrus, dysexecutivecorrelatessupramarginal of behaviourEF, gyri including and (DB). angular left Red gyri. regionsFigurepremotorUnique 2indicate neural andNeural supplemental correlates unique correlates neural of DB motor of are correlatesexecutiveregions,represented left functionof inlateral EF, teal, including (EF) orbital and and include left cortex and premotordysexecutivebilateralregions of middle and the behaviour supplemental right temporal frontal (DB). gyrus, pole, motor Red right Executive dysfunction on cognitive tests regions,regionssupramarginalorbitofrontal indicate left lateral cortex gyri unique orbitaland andangular neural bilateral cortex gyri. and bilateralcorrelatesUniqueanterior neural middlecingulate of EF, correlates temporal including cortex. ofgyrus, Finally, left DB are areas represented in teal, and include supramarginalpremotorof convergence and supplemental gyri between and angular EF andmotor gyri. DB are Dysexecutive behaviour Uniqueregions,regionsin yellow, neural of left and the lateralcorrelates right consist orbital frontal of of the cortex pole,DB majority are right and representedbilateralorbitofrontalof dorsolateral middle in cortex teal, prefrontal temporal and includebilateral cortex,gyrus, regionssupramarginalanteriorfrontal midline, of cingulate the right gyri regions cortex. and frontalangular and Finally, pole, bilateral gyri. right areas Both executive dysfunction of convergence between EF and DB are orbitofrontalUniqueorbitofrontal neural cortex cortex. correlates and Cortical bilateral of DB thickness are and dysexecutive behaviour anteriorrepresentedinof yellow, the regions cingulate and in consistindicatedteal, cortex. and of includeabove the Finally, majority have areas ofregionsofbeen convergence dorsolateral demonstrated of the right prefrontal between tofrontal be EF cortex, signi pole, andficantly DB right are inorbitofrontalfrontalassociated yellow, midline, and with cortex consist regions EF or and ofDB, and bilateral the respectively, bilateral majority ofanteriororbitofrontaland dorsolateral corrected cingulate cortex. for prefrontal multiplecortex. Cortical cortex, Finally, thickness areas frontalofcomparisons. convergencetheregions midline, indicated regionsbetween and above EF bilateraland have DB are fi orbitofrontalinbeen yellow, demonstrated and cortex. consist to Cortical of be the signi thickness majoritycantly ofofassociated the dorsolateral regions with indicated prefrontal EF or DB, above cortex, respectively, have beenfrontaland corrected demonstrated midline, for regions multiple to be and signi bilateralficantly associatedorbitofrontalcomparisons. with cortex. EF or Cortical DB, respectively, thickness andof the corrected regions for indicated multiple above have comparisons.been demonstrated to be significantly associated with EF or DB, respectively, and corrected for multiple comparisons.

45 Gansler et al (2017) JNNP

went up. As the proportion of bvFTD participants in low and Table 5 Binary logistic regression predicting bvFTD and CBS high DB groups was 27% and 77%, the severity level of DB diagnostic status from dysexecutive behaviour, executive function could explain the absence of a systematic relationship with EF. and visuo-motor function at step 3 Predictor B Wald χ2 p Value went up. As the proportion of bvFTD participants in low and Table 5 Binary logistic regression predicting bvFTD and CBS highDISCUSSION DB groups was 27% and 77%, the severity level of DB diagnostic status from dysexecutive behaviour, executive function EF and DB measures were found to have construct validity by Executive function 0.55 11.05 <0.01 wentcould up. explain As the the proportion absence of aof systematic bvFTD participants relationship in with low EF. and and visuo-motor function at step 3 virtue of being distinct but related classes of data by latent vari- TableDysexecutive 5 Binary behaviour logistic regression0.09 predicting 20.97 bvFTD and CBS <0.001 high DB groups was 27% and 77%, the severity level of DB diagnostic status from dysexecutive behaviour, executive2 function able factor analysis, that these constructs were associated with PredictorVisuo motor performanceB 0.62Wald 17.74 χ p<0.001 Value couldDISCUSSION explain the absence of a systematic relationship with EF. and visuo-motor function at step 3 wentconverging up. As and the divergingproportion neural of bvFTD correlates participants in the in prefrontal low and Table 5 Binary logistic regression predicting bvFTD and CBS ExecutiveBinary logistic function regression predicting 0.55 bvFTD and PPA 11.05 diagnostic status <0.01 from EFhighcortex, and DB that DB groups measures the measures was were27% displayed foundand 77%, to criterion have the constructseverity validity level validity as signi of DB byfi- Predictordiagnostic status from dysexecutiveB behaviour,Wald executiveχ2 functionp Value Dysexecutivedysexecutive behaviour behaviour, executive function0.09 and visual 20.97 confrontation <0.001 DISCUSSIONvirtuecouldcantly explain of worsened being the distinct EF absence and but DB of related a was systematic identi classesfied relationship of in data bvFTD by latent with and EF. vari- that andnaming visuo-motor at step 3 function at step 3 able factor analysis, that these constructs were associated with ExecutiveVisuo motor function performance 0.55 0.62 11.05 17.74 <0.01<0.001 EFthere and was DB incremental measures were utility found in their to have conjoint construct usage validityat low, but by 2 virtueconverging of being and distinct diverging but related neural classes correlates of data in the by latent prefrontal vari- DysexecutivePredictorExecutive function behaviour B0.090.44 20.97Wald 4.92 χ <0.001p<0.05 Value DISCUSSIONnot high, levels of DB. A counterintuitive finding is that diagno- Binary logistic regression predicting bvFTD and PPA diagnostic status from ablecortex, factor that analysis, the measures that these displayed constructs criterion were validity associated as signi withfi- VisuoDysexecutive motor performance behaviour 0.620.06 17.74 8.36 <0.001<0.01 EFsis and/or and DB severity measures level, were and found years to since have onset construct affected validity the rela- by Executivedysexecutive function behaviour, executive function 0.55 and visual 11.05 confrontation <0.01 convergingcantly worsened and diverging EF and DB neural was identi correlatesfied in in bvFTD the prefrontal and that namingConfrontation at step naming 3 0.15 14.74 <0.001 virtuetionship of of being EF anddistinct DB, but which related was classes systematic of data in PPA by latent and CBS, vari- BinaryDysexecutive logistic behaviour regression predicting0.09 bvFTD and PPA 20.97 diagnostic status <0.001 from cortex,there was that incremental the measures utility displayed in their criterion conjoint validity usage at as low, signi butfi- ableand non-systematicfactor analysis, inthat bvFTD. these constructs While EF were and associated DB modalities with dysexecutiveVisuo motor performance behaviour, executive function 0.62 and visual 17.74 confrontation<0.001 cantlynot high, worsened levels of EF DB. and A DBcounterintuitive was identifiedfinding in bvFTD is that and diagno- that Executive function 0.44 4.92 <0.05 convergingwere useful and in characterising diverging neural and identifyingcorrelates in the the dysexecutive prefrontal naming at step 3 theresis and/or was incremental severity level, utility and yearsin their since conjoint onset usage affected at thelow, rela- but BinaryDysexecutive logistic behaviour regression predicting0.06 bvFTD and PPA 8.36 diagnostic status<0.01 from cortex,syndrome, that DB the was measures the more displayed robust modality. criterion validityThe loss as of signi socialfi- nottionship high, of levels EF and of DB. DB, A which counterintuitive was systematicfinding in isPPA that and diagno- CBS, ExecutivedysexecutiveConfrontationTo better function understandnaming behaviour, executive DB and0.44 function0.15 EF contribution and visual 14.74 4.92 confrontation to classi<0.05fi<0.001cation, cantlycognition worsened in bvFTD EF and may DB have was a detrimental identified in effect bvFTD on and EF data that and non-systematic in bvFTD. While EF and DB modalities theirDysexecutivenaming relationship at step behaviour 3 was evaluated0.06 by group using 8.36 linear regression<0.01 sistherecollection and/or was severityincrementalrequiring level, motivation utility and years in and their since cooperation conjoint onsetaffected usage between at the low, exam- rela- but were useful in characterising and identifying the dysexecutive withConfrontation curve estimation.naming The relation0.15 of data 14.74 modalities<0.001 was not tionshipnotiner high, and of examinee, levels EF and of DB. but DB, Anot which counterintuitive on DB was data systematic collectionfinding in PPA fromis that and a diagno- partici- CBS, Executive function 2 0.44 4.92 <0.05 2 syndrome, DB was the more robust modality. The loss of social systematic for bvFTD (r =0.01), but was for PPA and CBS (r s andsispating and/or non-systematic caregiver. severity level, in bvFTD. and years While since EF onset and affected DB modalities the rela- DysexecutiveTo better behaviour understand DB and0.06 EF contribution 8.36 to classifi<0.01cation, cognition in bvFTD may have a detrimental effect on EF data of 0.37 and 0.32, respectively) who had lower levels of DB. weretionshipThe useful dysexecutive of EF in and characterising DB, syndrome which andof was disordered identifying systematic supervisory inthe PPA dysexecutive and atten- CBS, theirConfrontation relationship naming was evaluated0.15 by group 14.74 using linear regression<0.001 collection requiring motivation and cooperation26 between exam- Visual inspection of the PPA and CBS plots indicated a u-shaped syndrome,andtion andnon-systematic working DB was memory the in more bvFTD. processes robust While modality.has EF cognitive and The DB loss and modalities of behav- social with curve estimation. The relation of data modalitiesfi was not iner and examinee, but1242737 not on DB data collection from a partici- relationshipTo better understand in which theDBand left EF side contribution was more robustto classi thancation, the cognitionwereioural useful manifestations in inbvFTD characterising may have and a detrimental can identifying be observed effect the dysexecutive with on EF frontal data systematic for bvFTD (r2=0.01), but was for PPA and CBS (r2 s pating caregiver. 28 theirright, relationship but indicating was at evaluated very high by levels group of using DB EF linear performances regression collectionsyndrome,lobe pathology requiring DB was or other the motivation more neuropsychiatric robust and cooperation modality. aetiologies. The between loss ofPatients exam- social of 0.37 and 0.32, respectively) who had lower levels of DB. The dysexecutive syndrome of disordered supervisory atten- withTo curve better estimation. understand The DB and relation EF contribution of data modalities to classi wasfication, not inercognition and examinee, in bvFTD but may not have on DB a detrimental data collection effect from on a EF partici- data Visual inspection of the2 PPA and CBS plots indicated a u-shaped2 tion and working memory processes26 has cognitive and behav- systematictheir relationship for bvFTD was (revaluated=0.01), by but group was for using PPA linear and regression CBS (r s patingcollection caregiver. requiring motivation and cooperation between exam- ofrelationship 0.37 and in 0.32, which respectively) the left side who was had more lower robust levels than of DB. the iouralThe manifestations dysexecutive syndrome1242737and of disordered can be observed supervisory with frontal atten- with curve estimation. The relation of data modalities was not iner and examinee, but not on DB data26 collection from28 a partici- right, but indicating at very2 high levels of DB EF performances2 lobe pathology or other neuropsychiatric aetiologies. Patients Visualsystematic inspection for bvFTD of the (r PPA=0.01), and CBS but was plots for indicated PPA and a CBSu-shaped (r s tionpating and caregiver. working memory processes has cognitive and behav- relationship in which the left side was more robust than the ioural manifestations1242737and can be observed with frontal of 0.37 and 0.32, respectively) who had lower levels of DB. The dysexecutive syndrome of disordered supervisory28 atten- right,Visual but inspection indicating ofthe at very PPA high and CBS levels plots of DB indicated EF performances a u-shaped lobetionand pathology working or memory other neuropsychiatric processes26 has aetiologies. cognitive andPatients behav- relationship in which the left side was more robust than the ioural manifestations1242737and can be observed with frontal right, but indicating at very high levels of DB EF performances lobe pathology or other neuropsychiatric aetiologies.28 Patients Summary Executive function and dysexecutive syndrome

§ For overview: Chapter 3 covers frontal lobe function

§ See also separate reading list

46